Quantifying CO2 reduction potential through efficiency improvements in food waste upcycling: a network DEA and meta-frontier decomposition approach

This study evaluates stage-specific inefficiencies and CO2 reduction potentials in food-waste upcycling into animal feed. Using 43 months of input–output data from two Japanese dry-feed producers, we apply a network data envelopment analysis with a hybrid link structure that reflects the nondiscretionary nature of residue collection and the discretionary use of residues in production. To address seasonal heterogeneity, we incorporate a meta-frontier decomposition that benchmarks each month against best-practice observations within the same seasonal group. The results reveal persistent inefficiencies across both collection and production stages. Substantial theoretical CO2 reduction potentials—primarily associated with heavy oil and diesel use—could be realized through closer alignment with these best-practice months. Sensitivity analyzes using alternative link specifications, seasonal definitions, and outlier exclusions confirm the robustness of these findings. The integrated framework helps identify stage-specific improvement opportunities and clarify how efficiency gains in upcycling systems may relate to CO2 reductions.