Frontier-based performance analysis models for supply chain management: State of the art and research directions

Per J. Agrell, Adel Hatami-Marbini

Research output: Contribution to journalArticlepeer-review

47 Citations (Scopus)


Effective supply chain management relies on information integration and implementation of best practice techniques across the chain. Supply chains are examples of complex multi-stage systems with temporal and causal interrelations, operating multi-input and multi-output production and services under utilization of fixed and variable resources. Acknowledging the lack of system's view, the need to identify system-wide and individual effects as well as incorporating a coherent set of performance metrics, the recent literature reports on an increasing, but yet limited, number of applications of frontier analysis models (e.g. DEA) for the performance assessment of supply chains or networks. The relevant models in this respect are multi-stage models with various assumptions on the intermediate outputs and inputs, enabling the derivation of metrics for technical and cost efficiencies for the system as well as the autonomous links. This paper reviews the state of the art in network DEA modeling, in particular two-stage models, along with a critical review of the advanced applications that are reported in terms of the consistency of the underlying assumptions and the results derived. Consolidating current work in this range using the unified notations and comparison of the properties of the presented models, the paper is closed with recommendations for future research in terms of both theory and application.

Original languageEnglish
Pages (from-to)567-583
Number of pages17
JournalComputers and Industrial Engineering
Issue number3
Early online date28 Feb 2013
Publication statusPublished - 1 Nov 2013
Externally publishedYes


Dive into the research topics of 'Frontier-based performance analysis models for supply chain management: State of the art and research directions'. Together they form a unique fingerprint.

Cite this