Relational Drivers and Technical Efficiency in Indonesian Banana Value Chains: A Multi-Actor Regression-DEA Analysis

Abstract

Background: Indonesia is the world’s third-largest banana producer, yet its value chains remain structurally fragmented, with pervasive information asymmetries and substantial technical inefficiencies among smallholder farmers. Research integrating relational governance determinants with frontier efficiency analysis across upstream, middle, and downstream actors remains limited. Methods: A cross-sectional survey comprising 251 respondents (237 farmers, 5 wholesalers, and 9 retailers) was conducted across Cianjur, Lumajang, and Lampung, of whom 194 provided valid complete cases (181 farmers, 5 wholesalers, and 8 retailers). Constructs used reflective multi-item scales adapted from established supply chain governance literature; validity and reliability were confirmed through confirmatory factor analysis (Cronbach’s α = 0.731–0.941; composite reliability = 0.746–0.940; average variance extracted = 0.348–0.546). Antecedents of supply chain coordination were tested using hierarchical multiple regression on the farmer subsample with complete data (n = 181), and relative technical efficiency across value chain actors was estimated using input-oriented Data Envelopment Analysis (DEA). Results: The comprehensive regression model explained 63.3% of the variance in supply chain coordination performance. Inclusivity (β = 0.251), commitment (β = 0.259), information quality (β = 0.233), and trust (β = 0.160) were significant positive predictors, and a parsimonious model identified commitment and trust as the strongest. DEA showed low mean technical efficiency among smallholder farmers (0.683) relative to retailers (0.934) and wholesalers (0.954), with only modest variation across the three regions. Conclusions: Relational governance mechanisms primarily drive farmer-level supply chain coordination, whereas structural position within the value chain shapes technical efficiency. These empirically validated determinants and efficiency benchmarks can inform targeted agricultural policy. No artificial intelligence system was developed or tested in this study; the results are offered only as candidate inputs that could support the future design of AI-enabled decision-support tools in food supply chain management.