Global nickel scrap trade network vulnerability: Risks of exposure, cooperation disruptions and policy barriers

Abstract

Amid escalating geopolitical tensions, the global nickel resource trade is facing mounting systemic risks. This study develops a network-based framework that integrates structural exposure risk indicators and structural stress testing based on extinction analysis to assess the vulnerability of the global scrap nickel trade network (GSNTN). Results reveal a dual-risk structure characterized by intensified direct exposure and increasing efficiency imbalance. Four simulation scenarios of cooperation disruptions and policy barriers indicate that, nations exhibiting high dependency and reachability but low constraint tend to act as high-intensity risk. In contrast, highly constrained nodes embedded in cohesive trade clusters are prone to becoming passive vulnerable receptors, forced to absorb concentrated systemic pressure. Notably, some low-trade value intermediary countries act as disruption amplifiers. The findings highlight the vulnerability of the GSNTN, emphasizing that major countries should strengthen cooperation and avoid conflicts to ensure the stable operation of the supply chain.

Graphical abstract

Keywords

scrap nickel trade network / critical material supply chains / network vulnerability / extinction analysis / stress testing

1 Introduction

Under the global vision of carbon neutrality, nickel has emerged as a critical material carrier for the energy transition due to its pivotal role in ternary batteries and hydrogen technology (Su et al., 2025). However, this strategic demand currently faces severe supply-side constraints. The IEA predicts a doubling of global nickel demand by 2030, yet the geographic distribution of primary resources remains highly concentrated, with production monopolies in a few nations like Indonesia trapping the global trade in a state of deep external dependency (Axt et al., 2025; Calvo and Valero, 2022). More critically, the rise of resource nationalism and intensifying geopolitical rivalry mean that non-technical disruptions, such as export bans and investment barriers, have significantly increased uncertainty in raw material acquisition (Su et al., 2023; Wang et al., 2022). Coupled with the high negative environmental externalities of primary mining, the shortage of nickel resources has evolved into a systemic security challenge conditioned by the international political environment (Chen et al., 2022). To circumvent the vulnerability of primary trade, establishing an alternative supply system centered on secondary resources has become a global consensus (Zhang et al., 2023). The recycling of scrap nickel not only possesses significant low-carbon advantages, reducing energy consumption by approximately 70% compared to primary production (Eckelman, 2010), but is also regarded as a strategic buffer against geopolitical risks due to the widespread distribution of “urban mines” (Coman et al., 2013). Driven by these factors, the global scrap nickel trade network has expanded rapidly, serving as a key nexus linking disparate resource endowments and balancing supply-demand mismatches (Tan et al., 2021). However, as national industrial chains become deeply embedded within this network, the excessive concentration and complex coupling of trade relationships have embedded new risk latencies. In the context of trade protectionism and unstable bilateral cooperation, this tightly interlocking structure can easily transform local policy shocks into systemic crises, rendering the scrap nickel trade network itself a new source of vulnerability.

Although existing studies have begun to examine the trade evolution patterns of global metal resources, covering sectors such as scrap steel and copper (Hu et al., 2020a; Hu et al., 2020b; Li and Pu, 2025), a systematic exploration of the dynamic response of the scrap nickel trade network under specific policy shocks remains lacking. Existing literature largely focuses on static descriptions of flows or environmental benefit assessments, often overlooking the nonlinear behaviors of network topological structures in response to cooperation disruptions and policy barriers. Incorporating vulnerability and systemic risk theories is crucial in this regard. Vulnerability theory emphasizes the propensity of a system to fail under structural stress (Ding and Ren, 2025), while the systemic risk perspective focuses on how the failure of local nodes, such as major importers and exporters, triggers the paralysis of overall functions through the cascading effects of the trade (Mei et al., 2024; Sprecher et al., 2015). Against the backdrop of frequent global geopolitical frictions, identifying key risk sources, analyzing transmission paths of trade disruptions, and quantifying the destructive power of various policy barriers, such as unilateral sanctions and export bans, have become urgent tasks for safeguarding national resource security and industrial chain resilience.

Consequently, this study aims to construct a comprehensive framework integrating endogenous exposure risk and dynamic stress testing to assess the systemic vulnerability of the global scrap nickel trade network under cooperation disruptions and policy barriers. The main contributions of this paper are threefold. First, it provides a refined characterization of endogenous structural dimensions. Transcending traditional trade value statistics, this study quantifies the inherent structural exposure characteristics of the network from three topological dimensions: direct dependency, indirect transmission, and efficiency imbalance. Second, based on extinction analysis theory, four types of typical policy shock scenarios are introduced for structural stress testing. By constructing a flow redistribution model, the study quantitatively simulates the destructive potential of different types of cooperation disruption events within the network. Third, the study identifies the “critical failure chains” and “amplifier nodes” most sensitive to policy shocks within the network. Based on empirical findings, targeted strategies for trade optimization and diversified substitution are proposed, providing decision support for economies formulating resource security policies in a volatile international environment.

The paper is structured as follows: Section 2 reviews the relevant literature. Section 3 presents the methodology and data. Section 4 reports empirical results, and the final section concludes with policy implications.

2 Literature review

The renewable nature of scrap nickel increasingly enhances its role within the global resource circulation system. The “production, consumption, recycling, reuse” cycle characterizing the regenerative nickel flow network is gradually evolving into a key arena for inter-state industrial security competition. Consequently, research on scrap nickel has coalesced around three core themes: recycling technologies, international trade network characteristics, and risk assessment.

Within the circular economy framework, scrap nickel, as a high-value and recyclable resource, has attracted extensive attention in recovery and reuse technologies. Early studies focused primarily on source identification and process optimization, such as extraction and separation of battery cathode materials and nickel-plating waste (Du et al., 2022; Xiao et al., 2024). With the massive return of spent batteries from electric vehicles and other sectors, scrap nickel recovery pathways have become increasingly complex and diversified, highlighting the importance of green recycling technologies for complex battery components (Roy et al., 2022). Furthermore, lifecycle assessments demonstrate that scrap nickel reuse significantly reduces water consumption and greenhouse gas emissions compared to primary nickel, serving as a cornerstone for low-carbon metallurgy (Bai et al., 2022; Kallitsis et al., 2022). At the national level, studies have examined differences in scrap nickel recycling policies, industrial development, and incentives across economies including China, India, Republic of Korea, Germany, and Denmark (Jung et al., 2025; Su et al., 2023; Tan et al., 2021; Zhou et al., 2022), exploring how policy mechanisms and technological advances can improve recovery rates and utilization efficiency. However, current research remains predominantly engineering- and environment-focused, with limited analysis from the perspectives of global circulation networks and geopolitical interactions to assess its systemic role in ecological security and resource strategy.

With increasing demand for scrap nickel recycling and lowering technological barriers, international circulation of scrap nickel has become more frequent. However, due to uneven global distribution of scrap nickel resources, varying national recycling capacities, and heterogeneous policy orientations and environmental standards, scrap nickel trade exhibits pronounced heterogeneity and instability (Chen et al., 2024; Su et al., 2025). For example, China, as the world’s largest stainless-steel producer, is heavily reliant on scrap nickel. Since the 2017 ban on ‘foreign waste’ imports and stricter regulations on recycled resource entry, global scrap nickel flows and patterns have significantly shifted (Schulz, 2020). Multiple studies indicate that these policies not only altered the global trade focus of scrap metals but also compelled some economies to enhance local recycling capacity and diversify trade networks (Zhou et al., 2022). Scrap nickel trade essentially forms a complex network system where countries or economies act as nodes and trade relations as edges (Wang et al., 2023). Complex network analysis can reveal structural characteristics behind trade relations (Ma et al., 2022; Su et al., 2025) and carbon emission pathways (Wang et al., 2023). It has been increasingly applied to global trade studies of primary nickel resources (Chen et al., 2024; Su et al., 2023; Wang et al., 2022), scrap copper (Hu et al., 2020a), chromium (Gao et al., 2022), oil and gas (Sun et al., 2023), scrap iron, aluminum, and cobalt (Liu and Müller, 2013; Zhou et al., 2022). However, literature applying complex network analysis specifically to scrap nickel trade remains limited. Zhou et al. preliminarily analyzed the network structure of scrap nickel trade using k-means clustering and centrality metrics (Zhou et al., 2022), while Hu et al. further explored the evolution of network community structures (Hu et al., 2020b).

Amid setbacks to multilateralism and escalating trade tensions (Wang et al., 2025a; Xue and Li, 2023), the global scrap nickel economy faces numerous uncertainties, including tariff policies, environmental standards, technical barriers, and sudden events such as geopolitical conflicts (Wang et al., 2025b; Yu et al., 2024). These factors may trigger trade disruptions and amplify systemic risks (Yue et al., 2024). Existing complex network studies often simulate cascading failures under extreme scenarios via node failures and path blockages, combined with simulations to assess overall system resilience (Bellè et al., 2022; Hu et al., 2023). This approach has been pivotal in analyzing primary resource security and energy crises, effectively identifying key node countries’ systemic influence on global resource flows (Ouyang et al., 2024; Wang et al., 2025b). However, such models typically assume shocks arise from natural disasters or market volatility, overlooking more frequent and complex disturbances driven by policy changes and climate shocks (Schenker and Osberghaus, 2025; Zhang et al., 2025b). In recycled resource systems like scrap nickel, policy factors often determine cross-border flow feasibility (Feng et al., 2022), including restrictive waste import regulations and export controls on specific recycled metals (Chang et al., 2023). Recycled resource networks exhibit higher sensitivity to policy shifts, with nonlinear structural reconfigurations manifesting as flow redirection, increased trade transfer costs, and path congestion, potentially causing secondary cascading effects among countries (Dou et al., 2024; Liang et al., 2022; Liu et al., 2018).

In summary, despite progress in scrap nickel research, three key gaps remain: (1) recycling studies focus on micro-level technical pathways, lacking systemic assessment of scrap nickel’s role in global resource security. (2) network analyses largely describe structure without revealing inherent exposure risk underlying network formation. (3) existing models lack the integration of policy-driven trade disruptions within a structural stress-testing framework, failing to capture the latent vulnerabilities and redistribution pressures inherent in the network topology.

3 Methodology

3.1 Network construction

In the analysis of global resource trade networks, existing research primarily relies on two categories of data sources. The first category consists of shipping data based on the automatic identification system (AIS), which is frequently employed to construct energy transport networks where ports serve as nodes and routes as links (Xu et al., 2023; Zhang et al., 2025a). While AIS data accurately delineate physical transport routes and port accessibility, they fall short in comprehensively capturing trade dependencies. Specifically, these data are often insufficient for tracking cross-border raw material flows, recycling trade, and implicit commercial linkages. In contrast, data derived from the UN Comtrade offer a more direct reflection of resource supply-demand patterns and factor mobility. Consequently, this source is extensively utilized in the analysis of raw material and scrap trade networks (Hu et al., 2021; Zhang et al., 2024). To mitigate potential statistical noise within the UN Comtrade database, such as the underreporting of informal trade or HS code misclassification, the academic community has established robust processing protocols. For instance, Zhou et al. proposed correcting missing trade values via price interpolation and excluding negligible transactions (e.g., those under 50 kg) to ensure data comparability and validity (Zhou et al., 2022). Furthermore, Hu et al. noted that global trade networks exhibit significant scale-free characteristics and the Matthew effect (Hu et al., 2023). Therefore, even the omission of certain marginal or informal trade data does not substantively alter the core topological structure of the network. This theoretical consensus provides the justification for utilizing this database for the macro-systemic risk analysis in this study.

Accordingly, this study selects bilateral trade data under HS code 7503 (Nickel waste and scrap) from 2005 to 2024 as the primary data set. Addressing the asymmetry in mirror statistics caused by discrepancies in national statistical methodologies, we adopted an “average smoothing and complementary filling” data cleaning strategy (Yu et al., 2024). Specifically, for inconsistencies in bilaterally reported data, we followed established literature conventions by averaging the two figures to smooth statistical errors. For data reported unilaterally, we retained the records based on the principle of maximum information. Consequently, following these preprocessing and screening steps, the GSNTN model was constructed, that is G^t=(V^t,E^t,W^t), here, t representing the year. V^t denotes the set of economies participating in scrap nickel trade, V^t=\{v_1,v_2,...,v_n\}, where n is the number of participating countries. E^t represents trade links between economies, defined as E^t=\{e_{ij}^t\mid i,j\in V^t\}, where a directed edge from exporter v_i to importer v_j exists if trade occurs between them. W^t is the weighted adjacency matrix, W^t=\{w_{ij}^t\mid (i,j)\in E^t\}, where w_{ij}^t denotes the export value of scrap nickel from economy v_i to v_j in year t, measured in trade value. If w_{ij}^t>0, a positive export relationship from v_i to v_j exists for year t.

3.2 Structural exposure risk indicators

Structural exposure risk (SER) refers to the potential susceptibility and propagation capacity of network nodes (countries) arising from their structural positions within the network (Zarghami and Dumrak, 2021). In supply chain risk management research, existing literature has widely pointed out that structural attributes play a foundational role in determining a network's disturbance resistance and risk tolerance boundaries, particularly in terms of structural exposure characteristics manifested in resource concentration, node dependency, and information transmission (Greening and Rutherford, 2011). As a resource possessing both strategic significance and recycling value, the scrap nickel trade faces external shocks, such as policy barriers and transport disruptions, which are rarely isolated events (Yao et al., 2024). Instead, their impact pathways are often shaped and amplified by the network’s internal structural features. These hidden risk points in the cross-border flow process include direct exposure (e.g., overreliance on a single country), indirect transmission (e.g., policy changes in transit countries), and efficiency imbalances (e.g., elongated trade routes and high switching costs) (Cantner and Rake, 2014). Therefore, effective identification of latent risks within the GSNTN requires integrating the coupling mechanisms between endogenous structural vulnerabilities and external disturbances into the analytical framework. This study first quantifies the structural exposure risk borne by the network, encompassing direct exposure risk (DER), indirect transfer risk (ITR), and efficiency imbalance risk (EIR), to identify potential vulnerabilities in the transnational GSNTN.

Specifically, DER reflects the vulnerability of an economy arising from the number of its direct trading partners within the GSNTN. While maintaining multiple direct connections can enhance trade diversification, it may also increase the number of pathways for external policy disruptions to propagate. This indicator is measured by node strength and the distribution of strengths, as Eq. (1) in Table 1. A higher DER value indicates a greater concentration of direct trade flows, implying that the economy is subject to higher immediate exposure, where disruptions in these primary connections would directly result in significant trade value losses. ITR captures the extent of potential indirect shocks to which a node is exposed via multi-tiered trade linkages in the network. In practice, transit trade, cross-border reprocessing, and regional trade structures of scrap nickel mean that policy or transportation changes in non-direct trade partners can still be transmitted indirectly to the target country. To capture this, ITR is measured by the extent to which a node can reach higher-order (≥ 2) reachable paths through its neighboring nodes. For a given node v_i, the number of unique nodes reachable via paths of length k is denoted as IT{R}_i^{(k)}, as Eq. (2). From a risk perspective, a higher ITR value does not necessarily equate to enhanced robustness. Conversely, in highly coupled trade networks characterized by limited path redundancy, a larger ITR indicates that the node is embedded in complex indirect dependencies. In the event of an intermediary node failure, shocks can be synchronously amplified and transmitted to the node via multiple indirect pathways, thereby exacerbating its potential vulnerability. EIR is utilized to identify whether a node is situated in structural voids of information flow or resource allocation, measured by the structural hole constraint coefficient (Burt, 2004; Song et al., 2020). A lower constraint value indicates that a trader connects otherwise unconnected markets or intermediaries, serving as a typical “brokerage-critical node”. If such a trader faces policy sanctions, credit defaults, or logistics disruptions, the entire multi-party trade pathway may collapse, with concentrated risk borne by that node. This constitutes a “high-return, high-risk” structural vulnerability. Conversely, a higher EIR signifies that trade connections are densely clustered within a cohesive group of partners, implying strong structural constraints. While this configuration restricts brokerage efficiency, it typically limits the scope of potential risk spillover. The calculation is as Eq. (3).

3.3 Stress test model

Drawing on “extinction analysis” from ecological network studies, which assesses the impact of the loss of critical nodes or relationships on overall system functionality (Foti et al., 2013; Hu et al., 2021), this study further constructs a model for the impact of network external shocks based on representative scenarios. It aims to characterize the scope of structural imbalances and the degree of potential systemic exposure within the GSNTN resulting from external policy shocks, subject to existing trade structure constraints. Rather than simulating the temporal evolution of shocks, this model focuses on the initial phase of disruption. It functions as a structural stress test by proportionally redistributing disrupted flows among existing partners, thereby capturing the worst-case vulnerability of the post-shock structure under conditions of short-run rigidity.

Then, design four representative external shock scenarios. These include unilateral cooperation disruption, bilateral cooperation disruption, import bans, and export bans, which correspond to common real-world forms of trade partnership ruptures and policy barrier interventions. These scenarios are designed to simulate the immediate structural shock effects triggered by the failure of local trade relationships or critical nodes, assuming no immediate network reconstruction occurs. The selection of these four representative scenarios is not an arbitrary simplification. On one hand, intensified great power competition has made import restrictions, export controls, bilateral decoupling, and the blockade of critical channels common policy instruments in strategic resource governance. Given that the scrap nickel trade is characterized by concentrated resource endowments and limited substitutability, its trade structure is highly sensitive to policy interventions. Consequently, a single shock can expose significant structural vulnerabilities within the network. On the other hand, from the theoretical perspective of vulnerability and systemic risk, identifying system weaknesses does not require an exhaustive enumeration of all shock pathways. Instead, potential critical nodes, bottleneck channels, and high-exposure areas within the network can be effectively identified through a set of representative extreme scenarios. The model parameters are defined as shown in Table 2.

3.3.1 Trade partnership ruptures

(1) Unilateral trade disruption (UTD). If countryj unilaterally halts imports of scrap nickel from country i, which is assumed to reallocate the restricted export value across its existing trade network, aiming to preserve overall export equilibrium. Specifically, the blocked export value does not evaporate from the system, rather, it is reallocated to other importing partners of country i based on weighted preferences embedded in the existing trade structure. This mechanism reflects the exporter’s tendency to stabilize its export value due to domestic processing pressures and foreign exchange revenue considerations (Hu et al., 2020a). Assuming the export reduction from i to j is w_{ij}^t, the redistributed portion is allocated proportionally across country i’s other trading partners. Let f_1^t(w_{ij}^t,k) denote the redistributed trade flow from i to partner k at time t, then the absolute impact of the disruption can be expressed as:

(4)f_1^t(w_{ij}^t,k)=w_{ij}^t\cdot \frac{w_{ik}^t}{{\displaystyle {\sum }_{\left(i,h\right)\in {\tilde{E}}^t,h\ne j}{w}_{ih}^t}}\text{.}

From Eq. (4), the redistributed trade flow depends on two factors. The numerator w_{ik}^t reflects country k relative share within country i\text{'}s existing trade, meaning that the original proportion of i\text{'}s exports destined for k determines how much of the blocked export to partner j is reallocated to k. The denominator normalizes across all alternative destinations to ensure total trade value conservation. Consequently, even if the absolute weights are scaled differently, such as using actual trade values or logarithmic transformations, the redistribution ratios remain unchanged (Only changes the linear proportions), which demonstrates the robustness of the model to alternative weighting schemes. The relative impact of the shock, that is, the extent to which country k is affected, can be expressed by Eq. (5). A higher value of g_1^t(w_{ij}^t,k) indicates that country k must absorb a greater share of the redistributed import trade value f_1^t(w_{ij}^t,k), thus requiring stronger processing or adjustment capacity.

(5)g_1^t(w_{ij}^t,k)=\frac{f_1^t\left(w_{ij}^t,k\right)}{{\displaystyle {\sum }_{(h,k)\in E^t}{w}_{hk}^t}}\text{，}

However, since importing countries differ in their processing capacities, when g_1^t(w_{ij}^t,k) exceeds a certain threshold, country k is unable to accommodate the excessive redistributed trade value f_1^t(w_{ij}^t,k). Therefore, the number of systemically impacted countries is defined as follows:

(6)e_1^t(w_{ij}^t,\lambda )={\sum }_{(i,k)\in V^t}{I}_1^t(k,\lambda )\text{，}{I}_1^t(k,\lambda )=\{\begin{array}{l}1,\mathrm{i}\mathrm{f}{g}_1^t(\cdot )\ge \lambda \\ 0,\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{w}\mathrm{i}\mathrm{s}\mathrm{e}\end{array}\text{，}

where I_1^t(k,\lambda ) is an indicator function, and \underset{(i,j)\in E^{[t]}}{\mathrm{m}\mathrm{a}\mathrm{x}}{e}_1^t(w_{ij}^t,\lambda ) represents the maximum impact of unilateral trade disruption at time t, a larger value of e_1^t(w_{ij}^t,\lambda ) indicates higher vulnerability in the bilateral trade relationship w_{ij}^t. The greater the proportion of such high-risk supply and demand links in the network, the more vulnerable the entire GSNTN becomes. \lambda is a threshold parameter ranging from 0 to 1, follows the approach of Hu et al. in studies of scrap plastics, scrap iron, and copper raw material trade networks (Hu et al., 2021; Hu et al., 2020a). It is treated as a relative capacity threshold (i.e., a proportional coefficient) that reflects the heterogeneity in countries’ ability to absorb trade shocks. \lambda is not a purely structural or static assumption, but rather reflects differences in countries’ demand scale, processing and smelting capacity, industrial chain completeness, and policy adjustment capability in actual economic activities. In the model, g_1^t(w_{ij}^t,k) measures the proportion of redistributed trade that country k must absorb relative to its normal import trade value after a shock. A higher \lambda therefore indicates stronger pressure-bearing and processing capacity, larger market demand, or a more developed industrial system. In other words, although behavioral variables such as market demand shifts, technological progress, or industrial policies are not explicitly included in the model, they are implicitly captured through countries’ resource handling capacity and parametrized within \lambda, which in turn determines whether a country is classified as a systemically affected node under a shock. Essentially, \lambda serves as a composite proxy that condenses multidimensional economic capabilities, allowing the model to partially reflect heterogeneity and dynamic changes in real-world economic systems within a structural analysis framework.

(2) Bilateral trade disruption (BTD). Bilateral scrap nickel trade between countries i and j is disrupted due to factors such as political conflict or trade friction, resulting in a complete breakdown of two-way trade flows. It is assumed that the country with a net export position will internally redistribute the lost trade value across its remaining trade partners based on the residual net export. If country i is the net exporter, i.e., w_{ij}^t-w_{ji}^t>0, then the disrupted trade value will be proportionally reallocated to its other export destinations. The absolute and relative impacts of the shock are thus defined as follows:

(7)f_2^t(w_{ij}^t,k)=(w_{ij}^t-w_{ji}^t)\cdot \frac{w_{ik}^t}{{\displaystyle {\sum }_{(i,h)\in E^t,h\ne j}{w}_{ih}^t}},

(8)g_2^t(w_{ij}^t,k)=\frac{f_2^t(\cdot )}{{\displaystyle {\sum }_{(h,k)\in E^t}{w}_{hk}^t}}.

Similarly to Eq. (6), the number of countries impacted in the GSNTN under this scenario is calculated as follows:

(9)e_2^t\left(w_{ij}^t,\lambda \right)={\sum }_{\left(i,k\right)\in V^t}{I}_2^t\left(k,\lambda \right),I_2^t(k,\lambda )=\{\begin{array}{l}1,\mathrm{i}\mathrm{f}{g}_2^t(\cdot )\ge \lambda \\ 0,\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{w}\mathrm{i}\mathrm{s}\mathrm{e}\end{array}.

3.3.2 Policy barrier interventions

(1) Import ban (IB). Consider a scenario where a country unilaterally imposes import restrictions on scrap nickel due to stricter environmental policies or resource security concerns. Unlike the previous two scenarios, this external shock is initiated by the importer, whose blockade of inbound scrap nickel forces exporters to seek alternative markets. Assume that at time t, country i fully restricts scrap nickel demand from all countries. Consequently, all countries previously exporting to i are compelled to reallocate their planned export trade values to their other trading partners, maintaining overall export stability (Hu et al., 2020a). Under the assumption of trade value preference, the absolute and relative impacts of this shock on exporterjcan be expressed as follows:

(10)f_3^t(i,j)=s_i^t(\mathrm{i}\mathrm{n})\cdot \frac{w_{ji}^t}{{\displaystyle {\sum }_{(k,i)\in V^t}{w}_{ki}^t}},

(11)g_3^t(i,j)=\frac{f_3^t\left(i,j\right)}{{\displaystyle {\sum }_{(j,k)\in E^t}{w}_{jk}^t}}.

Since exporters often face rigid domestic pressures to process scrap nickel, their adjustment behavior tends to stabilize export trade values by rapidly redirecting the restricted share to existing trade channels. This can cause abnormal trade concentration or structural bottlenecks along original links. Similar to Eq. (6), the number of countries impacted in the GSNTN under this scenario is calculated as follows:

(12)e_3^t=\sum I_3^t\left(j,\lambda \right),I_3^t(j,\lambda )=\{\begin{array}{l}1,\mathrm{i}\mathrm{f}{g}_3^t\left(i,j\right)\ge \lambda \\ 0,\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{w}\mathrm{i}\mathrm{s}\mathrm{e}\end{array}.

(2) Export ban (EB). A country unilaterally prohibiting scrap nickel exports is common in cases of strategic stockpiling of critical raw materials or efforts to rebalance external dependencies. For example, since 2020, Indonesia has reinstated a comprehensive ban on nickel ore exports, posing a direct threat to downstream countries in the network. Assume that at time t, country iii imposes an export ban on scrap nickel, halting all outbound shipments. To mitigate supply-side disruptions, importers affected by this shock seek alternative suppliers within the existing trade network. Under the assumption of stable total import trade values, the affected country j redistributes its former import share from country i proportionally according to the trade weights of other existing suppliers. The absolute and relative impacts of this disruption on exporter j can be expressed as:

(13)f_4^t(i,j)=s_i^t(\mathrm{o}\mathrm{u}\mathrm{t})\cdot \frac{w_{ij}^t}{{\displaystyle {\sum }_{(i,k)\in E^t}{w}_{ik}^t}},

(14)g_4^t(i,j)=\frac{f_4^t\left(i,j\right)}{{\displaystyle {\sum }_{(h,j)\in E^t}{w}_{hj}^t}}.

This mechanism reflects the adaptive resilience of the GSNTN. When a core supply node fails, the system’s ability to effectively adjust through redundant pathways and maintain local or overall stability is key to assessing its vulnerability level. Similar to Eq. (6), the number of countries impacted in the GSNTN under this scenario is calculated as follows:

(15)e_4^t=\sum I_4^t\left(j,\lambda \right),I_4^t(j,\lambda )=\{\begin{array}{l}1,\mathrm{i}\mathrm{f}{g}_4^t\left(i,j\right)\ge \lambda \\ 0,\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{r}\mathrm{w}\mathrm{i}\mathrm{s}\mathrm{e}\end{array}.

4 Empirical results analysis

4.1 SER analysis

Figure 1(a) depicts the annual average evolution of DER and EIRin the GSNTN from 2005 to 2024. Overall, DER remained relatively stable between 2005 and 2017, averaging approximately 2.1 × 10⁴, indicating that most countries maintained limited and concentrated direct trade links in scrap nickel. However, since 2018, DER has risen significantly, exhibiting a sharp upward trend after 2020 and reaching nearly 3.1×10⁴ by 2024. This suggests a marked increase in the direct connectivity of nodes to external markets within the GSNTN, resulting in a higher degree of network coupling. Consequently, the system has become more sensitive to policy or transportation disruptions in individual economies, significantly elevating exposure risk. In contrast, EIR shows a gradual decline, dropping from an average of about 0.43 in 2005 to 0.41 in 2024. This indicates an overall decline in the network's structural hole constraint. While this evolution enhances resource circulation efficiency and cross-regional allocation capabilities, the expansion of structural holes implies that more nodes are embedded in critical intermediary positions. Consequently, local failures are more prone to amplification and spillover, thereby elevating the network’s level of systemic exposure.

Figure 1(b) further validates exposure risks through power-law characteristics. In four representative years, the log-log distributions fit power-law functions well, with slopes α ranging from –2.27 to –2.98. This indicates pronounced network heterogeneity: a few nodes bear a disproportionate number of direct trade connections, forming “super nodes”. Although this enhances local efficiency, it simultaneously creates potential single points of failure that can trigger cascading reactions during shocks. Figure 1(c)–1(d) reveal the top ten countries ranked by EIR and DER values in 2024. In terms of EIR, countries such as Austria, India, the UK, Finland, and China exhibit low structural constraints, indicating that they span multiple mutually disconnected trade clusters. These nations occupy typical intermediary positions and are characterized as efficiency-imbalance nodes. Conversely, within the DER dimension, Germany, the UK, Saudi Arabia, the USA, and India depend heavily on a limited set of trade partners and lack diversification mechanisms, representing highly exposed nodes. In summary, the GSNTN exhibits a dual evolution trend of increasing structural concentration and decreasing path redundancy. The concurrent intensification of DER and EIR highlights critical vulnerabilities in the network when facing external shocks.

Figure 2 illustrates the characteristics and dynamic evolution of indirect transfer risks at various path lengths within the GSNTN from 2005 to 2024. As shown in Fig. 2(a), countries’ exposure levels across paths of order 1 to 5 exhibit significant temporal fluctuations. Notably, the second-order paths consistently display the highest mean and distribution breadth, with average values of 33.2 and 33.3 in 2018 and 2024, respectively, indicating that most countries are indirectly connected to a broader range of scrap nickel sources through at least one intermediary, forming longer chains and more efficient risk transmission pathways. The third-order paths ITR³ have rapidly increased since 2011, reaching a mean of 17.3 in 2018, highlighting that, amid deepening global multilateral trade, risks propagate across regions via deeper path layers, substantially intensifying the system’s sensitivity to external shocks. In contrast, ITR⁴ and ITR⁵ remain at relatively low levels throughout, suggesting that high-order paths exist only among a few countries, and the depth of risk embedding has yet to fully materialize. Figure 2(b) presents the kernel density distribution of weighted shortest path lengths for each year, reflecting the overall compactness of indirect exposure. The results show that the main density peak for most years clusters below 0.05, demonstrating the clear small-world properties of the GSNTN: despite sparse direct connections, efficient indirect transmission structures persist. Since 2016, the mean path length has steadily shortened, dropping to 0.014 by 2024, indicating the network’s evolution toward a flatter and faster transmission structure. This accelerates cross-border indirect risk propagation and further amplifies systemic exposure.

Figures 2(c)–2(d) reveal changes in countries’ exposure patterns across multiple path orders between 2005 and 2024. In 2005, countries such as Mexico, Finland, South Korea, Spain, Netherlands, Belgium and China ranked high in ITR² and ITR³ exposure, this indicates that these economies are situated at the intersection of multiple cross-regional trade paths. Consequently, local upstream or downstream perturbations can readily propagate through them to the wider network, acting as structural “bridges” for cross-regional risk transmission. By 2024, emerging economies including Ireland, Switzerland, Poland, Turkey, China, Mexico, South Korea, and Canada have risen in prominence along ITR² and ITR³ dimensions, indicating a restructuring of indirect pathways within the GSNTN. This shift does not primarily stem from these nations actively functioning as “relay hubs”, but rather reflects a global reconfiguration of trade relations. As traditional core ties weaken, emerging recycling, reprocessing, and transit nodes are increasingly embedded into cross-border pathways. Consequently, formerly peripheral nodes are integrated into higher-order trade channels, transforming them into “path amplifiers”. Concurrently, the spatial reshaping of network structure and overall compression of path lengths enable risk signals to diffuse faster and farther within the multilateral trade system.

4.2 UTD analysis

To bridge the analytical gap between structural exposure metrics and shock simulation results, we establish a “structure-shock role” mapping framework. Within this framework, nodes or corridors that trigger the most extensive system-wide stress, characterized by high rankings in both the number of affected nations g_x^t and system stress intensity D(g_x^t), are identified as potential risk sources. Nodes prone to bearing supra-threshold loads during flow redistribution are classified as vulnerable receptors (indicated by the frequency of observing e_x^t and D(e_x^t)). Finally, nodes that possess high reachability across multi-order paths yet exhibit low structural constraints are identified as structural amplifiers, as they tend to “transmit rather than absorb” pressure. Figures 3 and 4 presents the top 20 disruptions under four scenarios in 2005 and 2024, respectively. As shown in Fig. 4(a), the disruption of the UK–USA trade corridor affects up to 31 countries, with 15 exceeding the threshold \lambda, demonstrating a significant spillover scale in the stress test. This impact is not solely due to the large trade volume, but primarily because both connected endpoints exhibit high DER (high load/dependency) and are positioned at the intersection of multi-order indirect paths (high ITR). This topological placement facilitates the spillover of redistribution pressure to numerous third-party nodes via existing channels. Concurrently, the associated hub nodes generally present low EIR, reflecting low structural constraint and strong brokerage, which makes them more prone to transmitting rather than absorbing shock pressure. Ultimately, this creates a “high load, high reachability, and low constraint” amplification structure.

Additionally, trade disruptions within the European Union, such as Germany to Switzerland (8 affected countries) and Germany to the UK (7 affected countries), reveal structural concentration risks inherent in regional integration systems. Germany’s industrial output forms a highly centralized layout within the multilateral network, disruption of its main import channels forces peripheral nodes to reconfigure import sources, thereby causing a broader reorganization of resource flows. These top 20 disruptions highlight how unilateral export dependencies amplify the vulnerability of the global scrap nickel supply security, demonstrating that despite the high reconfigurability of scrap nickel, the network structure still contains significant fragile cores.

Figures 6(a)–7(a) show that the UK→USA channel disruption primarily impacts central and western European and pan-European countries (e.g., Austria, Sweden, Netherlands), as well as emerging Eurasian nodes, reflecting its role as a bridging structure for resource redistribution within the EU. The disruption significantly increases reconfiguration costs and delay risks, with an impact magnitude of 7.19. The USA → Canada channel serves dual functions as a North American internal circulation hub and an Asia-Pacific re-export platform, its disruption affects strategic resource access for advanced manufacturing regions such as Japan and Singapore, exposing critical supply vulnerabilities in the global high-value chain, with an impact magnitude of 4.69. Meanwhile, the disruption of the Saudi Arabia → India route concentrates impact on Asian economies including China, South Korea, and Malaysia, indicating India’s pivotal role as a subregional transit and risk-sharing hub in the South Asia → East Asia recycled resource chain. Its disruption triggers a high-order pressure response characterized by bridge failure, regional stagnation, and multi-point amplification, with a total impact magnitude reaching 9.85.

4.3 BTD analysis

Figures 4(b) and 6(b), Figs. 3(b) and 5(b) illustrate that in the bilateral trade disruption simulations in 2024 and 2005, the top 20 critical trade pairs reveal the GSNTN’s pronounced path dependency and structural coupling, which drive systemic vulnerability. Risk exposure manifests primarily in the number of affected countries and the heterogeneity of system responses. Notably, the disruption of the UK ↔ USA, Saudi Arabia ↔ India, and Germany ↔ Sweden corridors caused systemic shocks in 14, 10, and 8 countries respectively, ranking high in both D ({\mathrm{g}}_2) and D ({\mathrm{e}}_2), thereby forming dense and uneven impact propagation patterns. Specifically, after the UK↔USA route disruption, the shock mainly affects to middle- and high-income countries or region such as Austria, France, the Netherlands, and Switzerland, as Fig. 7(b).This disruption affected 14 countries with an impact magnitude of 6.49, with high ITR and low EIR, these nations are embedded in multi-stage transit paths, rendering them prone to absorbing redistribution pressures and amplifying systemic impact. In contrast, the disruption of the Saudi Arabia–India corridor affected East Asian and Middle Eastern economies, including China, South Korea, and Malaysia. The peak impact magnitude of 9.85 highlights India’s status as a sub-regional hub within the South Asia–East Asia recycling resource chain, characterized by both high DER and high ITR.

Although the Germany–Sweden route affected fewer countries (8), the impacted nations, such as Denmark, Poland, and Italy are key central and eastern European nodes, indicating the corridor’s role as a crucial regional transmission bridge. Its disruption would significantly extend regional trade reconfiguration paths, triggering a “regional coupling weakening” effect. Additionally, the USA–Japan corridor, while directly affecting only 5 countries, influences 26 nations overall, including Canada, Mexico, and Chile in the Americas and Asia-Pacific. This reflects a bilateral dependency with a “high concentration–high output” structure. The relatively of high D({\mathrm{g}}_2) values and D({\mathrm{e}}_2)values reveal this corridor’s potential for “small-scale, high-intensity” risk amplification under crisis conditions.

4.4 IB analysis

Figure 4(c) presents the top 20 shocks under the IB scenario. The import disruptions by key recycled resource-importing countries such as the Netherlands, Germany, USA, and India generate the most significant systemic impacts, reflecting their critical roles as major receiving nodes in the GSNTN. Specifically, the Netherlands’ import ban causes up to 20 countries to face export disruptions, with an impact magnitude reaching 9.97. This indicates that the Netherlands not only connects diverse input sources, such as Belgium and Chile, but also serves as a hub for multi-node resource redistribution within the network. Its demand disruption would substantially amplify substitution costs for neighboring countries, causing localized disorder in the resource flow network. When Germany enacts an import ban, 33 countries are affected, marking the widest impact among all scenarios, with the highest impact magnitude of 15.62. This underscores Germany’s position as a central concentration point for recycled resources in Central Europe, characterized by strong network nesting and structural coupling. Its import interruption significantly amplifies impacts on regional countries including Austria, the Czech, France, and Poland, further confirming Germany’s role as a systemic core in regional resource allocation and risk transmission.

If the USA imposes an import ban, it triggers a system-wide affecting 28 countries across North America, Asia-Pacific, and Europe, such as Japan, Mexico, France, and South Korea, with a total impact magnitude of 14.33. This reflects the USA’s strong penetration and reverse dependency in the global high-value recycled metals market. In this scenario, the USA functions not only as a consumption endpoint but also as an absorption hub stabilizing the network’s coupling, its failure would cause widespread re-export imbalances and trade disruptions. Additionally, an import ban by India produces cross-regional shock effects impacting 19 countries, including China, Japan, the UAE, and Malaysia, which are key manufacturing and transit economies in Asia. The impact magnitude reaches 13.13, illustrating India’s dual role as a “demand bridge” and “transit amplifier” within the South-east Asia regional chain. Its disruption would break the systemic transmission chain and destabilize regional resource redistribution functions.

4.5 EB analysis

Figures 4(d) and 6(d) illustrate the top 20 shocks under the export ban scenario. Key exporting countries such as Germany, USA, UK, and Japan exhibit significant systemic impacts when their exports are interrupted, highlighting their roles as critical “export hubs” in GSNTN. Export flow disruptions from these countries trigger widespread cross-regional resource imbalances. Specifically, Germany’s export ban affects the normal imports of 16 countries, with an impact magnitude 6.66, indicating both considerable intensity and scope. Affected countries include industrial economies in central and western Europe such as Austria, France, Italy, Poland, and Sweden. This underscores Germany’s establishment of a stable recycled resource supply network within Europe, where any export disruption would undermine multi-node redundancy and severely threaten the security of the EU’s circular materials trade. Similarly, if USA enforces an export ban, 16 countries would suffer systemic impacts, with the highest impact magnitude of 7.16 among all countries, indicating the strongest shock intensity. The affected nations span Canada, Japan, Mexico, Singapore, and other global high-tech manufacturing and re-export hubs, reflecting the USA’s pivotal role as a resource redistribution platform connecting the Americas, Asia-Pacific, and Europe. An export halt would trigger multi-point supply-demand mismatches and block resource allocation in the global high-value chain.

If UK implements an export ban, 15 countries would experience systemic shocks. Austria, the Netherlands, Switzerland, and Thailand, as long-term export recipients, form diversified export paths linking pan-Europe and Asia, creating a cross-regional flow and multi-point penetration trade network. The disruption would significantly impair resource regulation capabilities between regions and destabilize the global scrap nickel trade. Japan’s export ban affects fewer countries or regions (8), but its impact targets, such as India, and the Philippines are situated in key South-east Asia manufacturing hubs, with an impact magnitude of 3.19. This demonstrates Japan’s core position in the regional scrap nickel circulation, where export interruptions increase the overall exposure risk for recycled nickel access in the Asia-Pacific region. Synthesizing the results across the four shock scenarios reveals that within the GSNTN, countries simultaneously exhibiting high DER (high load and dependency), high ITR (strong multi-order reachability), and low EIR (weak structural constraint) are predisposed to act as risk sources or amplifiers during systemic shocks. Conversely, nodes possessing both high DER and high EIR, due to their embedding in highly closed trade relationships, are more susceptible to absorbing concentrated pressure, thereby manifesting as passive structurally vulnerable nodes.

Overall, by comparing Fig. 3 and Fig. 5 as well as Fig. 4 and Fig. 6, it is observed that a high degree of consistency between the number of countries reached by each risk source and the actual impact imposed on these countries, indicating a strong coupling between the shock affect scale and the resulting damage mechanisms within the network. Notably, both the shock range and the impact intensity in 2024 are substantially higher than those in 2005, suggesting that the expansion of the GSNTN, increasing structural complexity, and deeper interdependencies among nodes have collectively heightened the system’s vulnerability to external disturbances. Meanwhile, under the current context of geopolitical tensions, frequent trade restrictions, and accelerating trade restructuring, the stability of traditional trade linkages has weakened, making shocks more likely to trigger cross-regional amplifying systemic risk exposure. Import–export bottleneck nodes consistently emerge as the most critical risk sources, with significantly broader shock ranges than marginal trade disruptions, underscoring the pivotal role of core nodes in maintaining the stability of the GSNTN.

To assess the robustness of the results, \lambda was further varied from 10% to 90% in increments of 10%, as Fig. 8. Overall, as \lambda increases, the number of affected countries significantly decreases, exhibiting a power-law decay trend, indicating the system’s heightened sensitivity to low-intensity disturbances. Meanwhile, risk shocks are highly concentrated in the top five events, with a steep early decline that reflects the dominance of a few high-intensity events in driving systemic risk pressure, illustrating the strong-node dominance property of the network structure. Comparing risk types, IB consistently trigger the broadest systemic vulnerability. The red pentagon line remains significantly higher than others across all \lambda values, indicating that policy controls by core importers have high spillover effects and easily initiate system-wide impact, causing many countries to lose export capabilities. EB rank second, especially in the moderate to low \lambda range, showing noticeable systemic influence and reflecting the structural amplification effect of supply interruptions at relay hubs. Regarding threshold sensitivity, differences among risk types are most pronounced when \lambda \le 0.3, with IB’s shock effect being especially prominent. For \lambda \ge 0.6, particularly between 0.7 and 0.9, except for IB which still exhibits some tail shocks, the other risk curves converge. This suggests the system’s exposure to high-intensity events concentrates on a few critical nodes, exhibiting a certain “strong-shock filtering” capacity. Moreover, the results indicate that variations in \lambda across the four scenarios only affect whether certain countries are identified as systemically impacted, while the overall shock propagation pattern and the macro-level systemic vulnerability of the network remain stable, demonstrating that the model outcomes are robust to the choice of \lambda.

5 Conclusions and implications

This study systematically investigates the vulnerability mechanisms of the GSNTN from 2005 to 2024. By integrating complex network analysis with extinction analysis models, this study constructed a structural stress testing framework to assess both exposure risk and the impact of policy shock shocks. The main conclusions are as follows:

First, the GSNTN exhibits a structure characterized by high coupling and structural concentration, resulting in a significant increase in endogenous exposure. Since 2018, the increase in DER signals the strengthening of core nodes, while the decline in EIR reflects the embedding of more nodes into critical intermediary roles. This marks a transition from a structure anchored by a few core ties to a highly coupled network dependent on specific intermediaries. Power-law distributions confirm that “super-nodes” carry disproportionate trade values, creating a trade-off where enhanced allocation efficiency comes at the cost of amplified systemic shock potential. It is recommended that for backbone trade channels with high dependency, alternative supply routes should be established, channel capacities be monitored in real time, and regular shock simulations and stress tests be conducted to enhance the resilience of key channels and prevent systemic impact triggered by single-point failures.

Second, the coexistence of extended indirect paths and small-world effects accelerates and deepens risk diffusion. Multi-level indirect transfer risks continue to intensify, alongside a sustained shortening of average network path length. This reflects an acceleration of hierarchical connectivity and “spatial flattening” in the scrap nickel trade system. The rise of emerging transit countries drives the “hub-ification” of peripheral nodes, exacerbating cross-regional risk shock, underscoring the urgent need to monitor and manage latent critical nodes.

Third, disruptions in key trade channels trigger significant systemic vulnerability. Simulation results indicate that interruptions in a few high-weight channels, such as UK ↔USA, Saudi Arabia ↔ India, and Germany ↔ Sweden, can rapidly impact more than a dozen countries. This reflects the highly concentrated dependency structure of the GSNTN. Due to the lack of effective redundant paths, the failure of these backbone channels can induce widespread trade imbalances and redistribution pressures through structural bottlenecks throughout the network, making overall system robustness heavily dependent on the normal operation of a small number of critical channels.

Fourth, trade barriers introduce substantial spillover risks. Scenarios involving import and export bans show that unilateral restrictions by core countries such as Germany, the USA, and the UK can affect more than 30 countries, triggering cascading amplifications that reflect systemic interdependencies. In a tightly coupled global network, these policies can lead to unintended exposure of otherwise low-risk nodes, underscoring the importance of structural sensitivity assessments in policymaking to prevent the inadvertent creation of system vulnerabilities.

Fifth, small relay countries play crucial systemic roles. Countries such as Belgium and Singapore, although not handling large trade values, often act as transit hubs within the network, performing functions of relaying, distributing, and re-exporting goods. Traditional metrics are insufficient to capture their systemic significance, making complex network analysis necessary to reveal their irreplaceable roles. Overlooking these functional nodes may create blind spots in emergency response planning and lead to an overestimation of the system’s overall resilience. For functional relay nodes, it is recommended to enhance their resilience through multi-tiered trade route optimization, diversification of trade dependencies, and increased network redundancy, ensuring that critical flows are maintained when major channels are disrupted and preventing single-point failures from amplifying systemic shocks.

This study still has several avenues for further refinement. This study still presents several avenues for further development. First, future research may incorporate the resilience triangle model to quantify the full trajectory from shock to degradation and recovery, thereby providing a more systematic assessment of the network’s capacity for shock absorption, loss magnitude, and recovery speed across different stages. Building on this foundation, the quantitative effects of various resilience-enhancing strategies, such as constructing redundant channels, promoting trade diversification, and strengthening regional coordination mechanisms, may be evaluated to identify the most cost-effective system optimization options. In addition, the long-term recovery capacity and adaptive capability of the trade network may be examined under evolving policy environments and changing market structures, contributing to a more comprehensive “shock, response, recovery and reinforcement” analytical framework and offering stronger theoretical support for resilience-building pathways. Beyond these conceptual and modeling extensions, further improvements can also be made with respect to data and methodological approaches. Future work may apply anomaly-detection techniques from machine learning and artificial intelligence, combined with industry statistics and firm-level reports, to correct for missing data, misreporting, and structural biases in trade statistics, thereby improving the accuracy and completeness of network construction. Moreover, the current shock-propagation model primarily captures direct impacts, while real-world trade risks often exhibit significant amplification effects, including trade contraction, downstream industrial dependence, midstream processing constraints, and the adjustment costs of alternative sourcing. Future research could develop a dynamic propagation mechanism that integrates supply substitution, demand responses, and inventory adjustments, and employ AI-assisted scenario generation and cascading failure methods to incorporate geopolitical conflicts, policy disruptions, and other tail-risk events into simulations. Intelligent optimization algorithms may further be used to identify potential substitution pathways and system recovery strategies.

References

[1] Axt M, Baldassarre B, Kirchherr J, (2025). Towards greater circularity in the hydrogen technology value chain. Ecological Economics, 236: 108679

[2] Bai Y, Zhang T, Zhai Y, Jia Y, Ren K, Hong J, (2022). Strategies for improving the environmental performance of nickel production in China: Insight into a life cycle assessment. Journal of Environmental Management, 312: 114949

[3] Bellè A, Zeng Z, Duval C, Sango M, Barros A, (2022). Modeling and vulnerability analysis of interdependent railway and power networks: Application to British test systems. Reliability Engineering & System Safety, 217: 108091

[4] Burt R S, (2004). Structural holes and good ideas. American Journal of Sociology, 110( 2): 349–399

[5] Calvo G, Valero A, (2022). Strategic mineral resources: Availability and future estimations for the renewable energy sector. Environmental Development, 41: 100640

[6] Cantner U, Rake B, (2014). International research networks in pharmaceuticals: Structure and dynamics. Research Policy, 43( 2): 333–348

[7] Chang L, Taghizadeh-Hesary F, Mohsin M, (2023). Role of mineral resources trade in renewable energy development. Renewable & Sustainable Energy Reviews, 181: 113321

[8] Chen K, Li Y, Linderman K, (2022). Supply network resilience learning: An exploratory data analytics study. Decision Sciences, 53( 1): 8–27

[9] Chen W, Jiang Y, Liu Z, (2024). Unveiling structural differentiation in the global nickel trade network: A product chain perspective. Journal of Geographical Sciences, 34( 4): 763–778

[10] Coman V, Robotin B, Ilea P, (2013). Nickel recovery/removal from industrial wastes: A review. Resources, Conservation and Recycling, 73: 229–238

[11] Ding C, Ren X, (2025). Evolution and vulnerability analysis of global photovoltaic industry chain trade pattern. Scientific Reports, 15( 1): 1–16

[12] Dou S, Zhu Y, Liu J, Xu D, (2024). The power of mineral: Shock of the global supply chain from resource nationalism. World Development, 184: 106758

[13] Du K, Ang E H, Wu X, Liu Y, (2022). Progresses in sustainable recycling technology of spent lithium-ion batteries. Energy & Environmental Materials, 5( 4): 1012–1036

[14] Eckelman M J, (2010). Facility-level energy and greenhouse gas life-cycle assessment of the global nickel industry. Resources, Conservation and Recycling, 54( 4): 256–266

[15] Feng P, Zhou X, Zhang D, Chen Z, Wang S, (2022). The impact of trade policy on global supply chain network equilibrium: A new perspective of product-market chain competition. Omega, 109: 102612

[16] Foti N J, Pauls S, Rockmore D N, (2013). Stability of the world trade web over time—An extinction analysis. Journal of Economic Dynamics & Control, 37( 9): 1889–1910

[17] Gao Z, Geng Y, Xiao S, Zhuang M, (2022). Mapping the global anthropogenic chromium cycle: Implications for resource efficiency and potential supply risk. Environmental Science & Technology, 56( 15): 10904–10915

[18] Greening P, Rutherford C, (2011). Disruptions and supply networks: a multi-level, multi-theoretical relational perspective. International Journal of Logistics Management, 22( 1): 104–126

[19] Hu X, Wang C, Lim M K, Bai X, Yao C, (2021). Evaluating waste and scrap trade risks in Belt and Road Initiative countries. Resources, Conservation and Recycling, 173: 105728

[20] Hu X, Wang C, Lim M K, Chen W Q, (2020a). Characteristics of the global copper raw materials and scrap trade systems and the policy impacts of China’s import ban. Ecological Economics, 172: 106626

[21] Hu X, Wang C, Lim M K, Chen W Q, Teng L, Wang P, Wang H, Zhang C, Yao C, Ghadimi P, (2023). Critical systemic risk sources in global lithium-ion battery supply networks: Static and dynamic network perspectives. Renewable & Sustainable Energy Reviews, 173: 113083

[22] Hu X, Wang C, Lim M K, Koh S L, (2020b). Characteristics and community evolution patterns of the international scrap metal trade. Journal of Cleaner Production, 243: 118576

[23] Jung S, An S, Park J, Park H S, Bi X, (2025). Assessment of a critical mineral recycling network: A case study on nickel recovery from production waste in Korean eco-industrial parks. Journal of Industrial Ecology, 29( 4): 1197–1207

[24] Kallitsis E, Korre A, Kelsall G H, (2022). Life cycle assessment of recycling options for automotive Li-ion battery packs. Journal of Cleaner Production, 371: 133636

[25] Li Y, Pu Y, (2025). Pattern evolution and dynamic formation mechanism of global scrap copper trade network: Based on temporal exponential random graph model. Ecological Economics, 236: 108664

[26] Liang C, Umar M, Ma F, Huynh T L, (2022). Climate policy uncertainty and world renewable energy index volatility forecasting. Technological Forecasting and Social Change, 182: 121810

[27] Liu G, Müller D B, (2013). Mapping the global journey of anthropogenic aluminum: A trade-linked multilevel material flow analysis. Environmental Science & Technology, 47( 20): 11873–11881

[28] Liu H, Tian Z, Huang A, Yang Z, (2018). Analysis of vulnerabilities in maritime supply chains. Reliability Engineering & System Safety, 169: 475–484

[29] Ma Y, Wang M, Li X, (2022). Analysis of the characteristics and stability of the global complex nickel ore trade network. Resources Policy, 79: 103089

[30] Mei Q, Qinyou H, Hu Y, Yang Y, Liu X, Huang Z, Wang P, (2024). Structural analysis and vulnerability assessment of the European LNG maritime supply chain network (2018–2020). Ocean and Coastal Management, 253: 107126

[31] Ouyang X, Liu L, Chen W, Wang C, Sun X, He C, Liu G, (2024). Systematic risks of the global lithium supply chain network: from static topological structures to cascading failure dynamics. Environmental Science & Technology, 58( 50): 22135–22147

[32] Roy J J, Rarotra S, Krikstolaityte V, Zhuoran K W, Cindy Y D I, Tan X Y, Carboni M, Meyer D, Yan Q, Srinivasan M, (2022). Green recycling methods to treat lithium-ion batteries E-waste: A circular approach to sustainability. Advanced Materials, 34( 25): 2103346

[33] Schenker O, Osberghaus D, (2025). International trade and the transmission of temperature shocks. Environmental and Resource Economics, 88( 4): 965–1007

[34] Schulz Y, (2020). Chinese engagement abroad in the scrap business. China Perspectives, 2020( 4): 49–57

[35] Song Y, Yang N, Zhang Y, Wang J, (2020). Do more structural holes lead to more risk propagation in R&D networks. Management Decision, 58( 1): 39–57

[36] Sprecher B, Daigo I, Murakami S, Kleijn R, Vos M, Kramer G J, (2015). Framework for resilience in material supply chains, with a case study from the 2010 rare earth crisis. Environmental Science & Technology, 49( 11): 6740–6750

[37] Su C, Geng Y, van Ewijk S, Borrion A, Zhang C, (2025). Uncovering the evolution of the global Nickel cycle and trade networks. Resources, Conservation and Recycling, 215: 108164

[38] Su C, Geng Y, Zeng X, Gao Z, Song X, (2023). Uncovering the features of nickel flows in China. Resources, Conservation and Recycling, 188: 106702

[39] Sun X, Wei Y, Jin Y, Song W, Li X, (2023). The evolution of structural resilience of global oil and gas resources trade network. Global Networks, 23( 2): 391–411

[40] Tan J, Wehde M V, Brønd F, Kalvig P, (2021). Traded metal scrap, traded alloying elements: A case study of Denmark and implications for circular economy. Resources, Conservation and Recycling, 168: 105242

[41] Wang F, Li K, Chen X, Zhang W, (2025a). Assessing supply chain risks for chip industry with LDA and multi-layer Bayesian network method. Frontiers of Engineering Management, 12( 4): 1037–1057

[42] Wang J, Tan X, Liu D, (2025b). Critical risks in an industry chain-based global lithium supply networks: Static structure and dynamic propagation. Process Safety and Environmental Protection, 198: 107137

[43] Wang M, Guo Y, Hu H, Ding S, (2023). Embodied carbon emission flow network analysis of the global nickel industry chain based on complex network. Sustainable Production and Consumption, 42: 380–391

[44] Wang X, Wang A, Zhong W, Zhu D, Wang C, (2022). Analysis of international nickel flow based on the industrial chain. Resources Policy, 77: 102729

[45] Xiao Y, Li L, He J, Sun Y, Lei Y, (2024). A metallurgical approach for separation and recovery of Cu, Cr, and Ni from electroplating sludge. Science of the Total Environment, 921: 171130

[46] Xu M, Deng W, Zhu Y, Lü L, (2023). Assessing and improving the structural robustness of global liner shipping system: A motif-based network science approach. Reliability Engineering & System Safety, 240: 109576

[47] Xue J, Li G, (2023). Balancing resilience and efficiency in supply chains: Roles of disruptive technologies under Industry 4.0. Frontiers of Engineering Management, 10( 1): 171–176

[48] Yao T, Li Z Y, Zhang Y J, (2024). Medium-and long-term sustainable supply approaches and strategies for essential and typical strategic resources in China. Frontiers of Engineering Management, 11( 3): 576–583

[49] Yu Y, Ma D, Wang Y, (2024). Structural resilience evolution and vulnerability assessment of semiconductor materials supply network in the global semiconductor industry. International Journal of Production Economics, 270: 109172

[50] Yue X, Mu D, Wang C, Ren H, Peng R, Du J, (2024). Critical risks in global supply networks: A static structure and dynamic propagation perspective. Reliability Engineering & System Safety, 242: 109728

[51] Zarghami S A, Dumrak J, (2021). Unearthing vulnerability of supply provision in logistics networks to the black swan events: Applications of entropy theory and network analysis. Reliability Engineering & System Safety, 215: 107798

[52] Zhang H, Cao H, Guo Y, (2024). The time-varying impact of geopolitical relations on rare earth trade networks: What is the role of China’s rare earth export restrictions. Technological Forecasting and Social Change, 206: 123550

[53] Zhang H, Liu G, Li J, Qiao D, Zhang S, Li T, Guo X, Liu M, (2023). Modeling the impact of nickel recycling from batteries on nickel demand during vehicle electrification in China from 2010 to 2050. Science of the Total Environment, 859: 159964

[54] Zhang L, Su W, Liao S, Wang S, (2025a). Enhancing energy security through multi-scale network analysis: robustness in global crude oil shipping–trade networks. Reliability Engineering & System Safety, 265: 111525

[55] Zhang M, Liu D, Shui X, Hu W, Zhan Y, (2025b). Examining the impact of trade tariffs on semiconductor firms’ environmental performance. International Journal of Production Economics, 281: 109528

[56] Zhou X, Zhang H, Zheng S, Xing W, (2022). The global recycling trade for twelve critical metals: Based on trade pattern and trade quality analysis. Sustainable Production and Consumption, 33: 831–845

RIGHTS & PERMISSIONS

Higher Education Press