Supply chain performance evaluation using a network data envelopment analysis model with bias-corrected estimates

We measure the performance of a real-world triadic supply chain (SC) network consisting of suppliers, 3PL, and retailers. Three-stage network slacks-based measure (NSBM) and multi-parametric bias correction generalised directional distance function (MPBC-NGDDF) approach are developed to deal with conventional and non-discretionary inputs, desirable and undesirable linking activities and outputs. The sample consists of 786 combinations of triad members. Using both NSBM and MPBC-NGGDF facilitates a sound SC performance evaluation and identification of optimal inputs and outputs. We also compare the NSBM and MPBC-NGDDF estimates against the network epsilon-based measure (NEBM) efficiencies. This comparative analysis reveals the weaknesses and strengths of each method in the context of a complex SC network. Identifying weaknesses and strengths draws on the appropriateness of optimal solutions and practical implications. The MPBC-NGDDF approach meets both appropriateness criteria. According to this approach, considerable reductions are needed in the number of employees used by suppliers, the production agility, the origin port distance from the suppliers, the destination port distance from the retailers, the time from placing an order to delivery to the retailers' premises, and distribution costs.