Allocation Policies to Fulfil Heterogeneous Service Requirements under Resource Pooling
研究了在资源池化模型中，供应商如何设计分配策略来满足多个买家不同的服务水平要求，提出了两种新策略并与现有策略比较，为不同惩罚结构和绩效评估周期下的策略选择提供指导。
ABSTRACT Designing effective settings for performance measures (e.g., fill rate) of a service‐level agreement (SLA) is challenging. This challenge is intensified when a firm adopts the pooling inventory model to allocate inventory/capacity to multiple buyers. Each buyer has its own service‐level contract outlining the required service level, the penalty structures, and the performance review period (PRP) length, which might not be the same as other buyers. This means the supplier requires an effective resource allocation policy whereby the different requirements of multiple buyers are integrated into a pooling model and capacities/inventories are allocated in the most effective way. Given a base‐stock replenishment policy and finite time horizon PRP, in this study we propose two new (anticipative) allocation policies—foresight linear programming (FLP) and two‐stage stochastic (TS)—and compare them with existing allocation policies. These allocation policies are developed for different penalty structures of linear, lump‐sum, hybrid, and no‐penalty settings. Results show that suppliers benefit from longer PRPs if linear or hybrid penalty structures are employed. We also find that when the length of PRP of buyers is not identical, TS is the recommending policy. Further, results provide a guideline for selecting the best resource allocation policy under various SLA terms, in particular, where buyers' PRP lengths are not identical.