Data-driven causal networks for pork supply chain risk analysis in Sichuan Province

The stability of critical agricultural supply chains is fundamental to global food security, yet their inherent complexity challenges traditional risk management approaches. This paper introduces and validates a data-driven causal network framework, employing the Peter and Clark Momentary Conditional Independence (PCMCI) algorithm, to systematically disentangle these complex dynamics. We demonstrate the framework's efficacy through its application to the pork supply chain of Sichuan Province, a microcosm of the world's largest pork market. The framework successfully identifies breeding sow inventory, piglet prices, and major disease shocks as the dominant drivers of pork price fluctuations. Crucially, it quantifies their causal pathways and lag structures, revealing that changes in breeding sow inventory, for instance, impact market prices with a potent two-month lead time-a critical signal for early-warning systems. Building upon these empirically grounded causal pathways, the framework provides evidence for transitioning from reactive to proactive risk management. This research provides not only deep insights into the pork market but, more broadly, a replicable methodological approach for analyzing and bolstering the resilience of complex agricultural supply chains globally.