Optimal production and inventory policies of priority and price-differentiated customers

Abstract

Many firms are exploring production and supply chain strategies when customers may be segmented into different classes based on service level or priority. Such segmentation can result in a more efficient production system as well as a better match between supply and demand. In this research, we analyze a system with customer classes 1 and 2, where customer class 1 has a higher priority of fulfillment than customer class 2 in the same period. We develop an optimal production and inventory strategy that rations current and future limited capacity between customer classes 1 and 2, through reserving inventory for the future and accepting orders now for future delivery when demand and production are general stochastic functions. We show that a modified order-up-to policy (S*, R i*, B i*) is optimal in each period. S* is the targeted inventory level after production at the beginning of the period; R ^1* represents the optimal inventory to be protected from being sold to both classes, and R ^2* is the additional amount of inventory to protect from class 2. B ^2* is the optimal amount of future capacity to make available to both classes through backlogging, and B ^1* is the additional backlogging amount for class 1. Computational analysis shows that the differentiation strategy can result in a significant profit improvement over a traditional inventory policy.

Keywords:

• Inventory policy
• tactical inventory
• customer differentiation
• threshold policies
• two-class
• priority differentiated
• price differentiated

Acknowledgements

The research was supported in part by NSF grants DMI-0245352 and DMI-0348532, Lucent Technologies through The Logistics Institute, The Logistics Institute-Asia Pacific, and The Center for Engineering Logistics and Distribution. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation or any other sponsor. The authors would also like to thank the reviewers, Associate Editor, and Editor-in-Chief for their contributions, which greatly improved the paper.

Notes

¹Each decision is equal to zero if the condition is never satisfied.

²If l indicates that orders must be filled by period t+l, and previously accepted orders must be filled before new orders are accepted, then the results in this paper hold as described for planned backlogs. However, for the version of the problem with specific and varying l _t , the assumption that previous orders are filled first may be too restrictive.