Inventory Allocation in the Presence of Service‐Level Agreements

• Published in: Production and operations management Vol. 27; no. 3; pp. 553 - 577
• Main Authors: Chen, Chun‐Miin (Jimmy), Thomas, Douglas J.
• Format: Journal Article
• Language: English
• Published: Los Angeles, CA SAGE Publications 01-03-2018; Blackwell Publishers Inc
• Subjects: Compliance; fill rate; Inventory; inventory allocation; Inventory management; Operations management; review horizon; Service level agreements; service‐level agreement; Supply chains; fill rate; service‐level agreement; inventory allocation; review horizon
• ISSN: 1059-1478; 1937-5956
• DOI: 10.1111/poms.12814

Retailers often use service‐level agreements (SLAs) to evaluate their supplier's performance. Based on an examination of 70 SLAs from practice, we conclude that in terms of evaluating fill rate, these SLAs vary in at least three key dimensions: (i) Supplier performance can be evaluated for each demand request or over some longer horizon, (ii) the acceptable fill rate can be 100% or something less than 100%, and (iii) the non‐compliance charge can be a flat fee, a per‐unit‐short fee, or both. For a supplier operating a periodic‐review inventory system and serving multiple retailers, each with an SLA, we investigate the impact of inventory allocation rules on SLA compliance and expected non‐compliance costs across the three dimensions of review horizon, fill rate target and non‐compliance cost structure. We derive several analytic results when retailers are symmetric. For the single‐period setting and a fill rate target of 100%, we characterize the optimal allocation rule and rank other rules common in practice and the literature. We also characterize the optimal rule for the single‐period setting when the fill rate target is less than 100% and the SLA specifies a flat‐fee non‐compliance charge. In the multiple‐period case, we characterize the optimal rule when the fill rate target is 100%. When the fill rate target is less than 100% and the performance review horizon is more than one period, we develop a simple heuristic that outperforms common allocation rules.