An integrated guaranteed- and stochastic-service approach to inventory optimization in supply chains

•We integrate the stochastic- and guaranteed-service inventory optimization models.•The new hybrid-service (HS) approach exploits the benefits of both pure approaches.•We develop an optimization procedure for serial supply chains.•The HS approach performs at least as well as the better of the two pure approaches.•In many settings of our numerical study it delivers additional cost savings.

Multi-echelon inventory optimization literature distinguishes stochastic- (SS) and guaranteed-service (GS) approaches as mutually exclusive frameworks. While the GS approach considers flexibility measures at the stages to deal with stockouts, the SS approach only relies on safety stock. Within a supply chain, flexibility levels might differ between stages rendering them appropriate candidates for one approach or the other. The existing approaches, however, require the selection of a single framework for the entire supply chain instead of a stage-wise choice. We develop an integrated hybrid-service (HS) approach which endogenously determines the overall cost-optimal approach for each stage and computes the required inventory levels. We present a dynamic programming optimization algorithm for serial supply chains that partitions the entire system into subchains of different types. From a numerical study we find that, besides implicitly choosing the better of the two pure frameworks, whose cost differences can be considerable, the HS approach enables additional pipeline and on-hand stock cost savings. We further identify drivers for the preferability of the HS approach.