A Causal-Driven Data Architecture for Block-chain Based Agricultural Supply Chains

Abstract

Blockchain technology has been adopted in agricultural supply chains primarily as an immutable record-keeping mechanism. This paper argues that such usage is architecturally insufficient and proposes a four-stage causal-analytic framework repositioning blockchain as a predictive analytics platform for seed traceability. The central contribution is a data-driven methodology for determining which supply chain variables should reside on-chain versus off-chain, based on causal relevance rather than regulatory compliance or storage cost. A 600-record corpus integrates a 500-record calibrated simulation of the Indian certified seed supply chain anchored to 12 published institutional benchmarks (ICAR, NSC, NABARD, SSCA, FAOSTAT) with a 100-record real-world operational dataset covering five Indian cities. Experiment 1 applies the PC causal graph discovery algorithm, retaining 34 edges and classifying 18 variables as on-chain and 1 as off-chain. Experiment 2 computes E-values for 56 on-chain paths; 40 are significant (p<0.05), with maximum trust weight w=1.000 (E-value=1.875). Building upon this verified data topology, the ongoing scope includes Experiment 3 to derive smart contract triggers via serial mediation, targeting a Germination Rate threshold associated with localized complaint risk, and Experiment 4 to construct the Seed Traceability Risk Score (STRS) to optimize feature verification costs. This completed ledger architecture delivers an 18-variable on-chain storage schema, demonstrating that causal structure determines the predictive value of a traceability block chain.