Explainable Analytics of Blockchain-Driven Green Innovation: A Multi-Factor Framework for Supply Chain Systems
Main article
Abstract
Blockchain technology application (BTA) is increasingly deployed in manufacturing supply chains as a digital infrastructure for traceability, cross-organisational coordination, and automated compliance, yet empirical evidence on how BTA converts into green innovation efficiency (GIE) remains fragmented and methodologically narrow. This study develops a multi-factor analytical framework that combines a theoretically grounded structural equation model with a gradient-boosting ensemble and SHAP-based explainable analytics to examine how BTA, supply chain integration (SCI), and three contextual factors — supply chain trust, task complexity, and a green digital learning orientation — jointly shape GIE. Survey data from 380 Singapore manufacturers distributed across seven industry sub-sectors were analysed with a two-stage procedure. Stage one estimates a theory-driven path model showing that SCI fully mediates the BTA–GIE relationship (indirect β = 0.049, 95% bootstrap CI [0.023, 0.078]), with trust amplifying and task complexity attenuating the BTA→SCI pathway. Stage two benchmarks six predictive model families, ranks feature importance through mean absolute SHAP values, and maps nonlinear partial-dependence responses. The stacked ensemble attains R² = 0.598 out-of-sample, outperforming the structural model by 13.7 percentage points and revealing a previously unobserved saturating effect of BTA above the mid-range of the scale. Heterogeneity analysis indicates that the BTA→GIE total effect is substantially larger in large firms, high-pollution sectors, and export-intensive firms. The study advances the literature by integrating confirmatory and exploratory analytics within a single explainable framework and offers managers a concrete, interpretable toolkit for prioritising digital-sustainability investments.
