Hierarchical integration in critical-mineral markets: spillover asymmetry and systemic risk in the energy transition
摘要
This study reveals that nine critical minerals underpinning the global energy transition form a hierarchically integrated commodity network, one that is simultaneously highly interconnected and persistently asymmetric in its shock transmission structure. Using annual price data spanning 1962–2020, and employing Diebold–Yilmaz connectedness indices, rolling-window analysis, saliency mapping, ablation studies, and Diebold–Mariano predictive tests, we find that nearly half of all forecast error variance is attributable to cross-mineral shocks, with connectedness surging above 70% after 2005 during the energy-transition era. Critically, copper and aluminium emerge as dominant, persistent net transmitters of shocks, while graphite, lithium, and silicon function as stable net receivers a directional asymmetry that contradicts symmetric financialization predictions and instead reflects supply-chain hierarchies rooted in market liquidity and geographic concentration. Silver operates as a volatile amplifier, exhibiting episodic swings between transmission and reception driven by its dual industrial and financial identity. Despite this extreme connectedness, cross-mineral information yields negligible incremental forecasting gains over univariate benchmarks, with no single mineral proving predictively indispensable once estimation instability is controlled through regularization. These findings imply that systemic vulnerabilities in critical-mineral markets are concentrated in a narrow transmission backbone, posing significant risks for energy-transition supply chains. For policymakers, the hierarchical structure suggests that supply-security interventions should target shock-originating and backbone nodes, particularly copper and aluminium, rather than treating the system as uniformly integrated, while the informational flatness underscores the limits of cross-mineral price signals for strategic stockpiling or early-warning systems.