Network inventory deployment for responsive fulfillment

Customer expectations on fulfillment responsiveness is growing with prevalent home delivery practices via e-retails or omnichannel operations. Tight responsiveness requirements, such as x-hour delivery, requires physical availability of inventory near demand locations. This raises a need for a broad and dense inventory network and a smart inventory deployment model. Although building such a decentralized network is financially unviable for most companies, the recently emerging Physical Internet can enable affordable access to a decentralized network by transforming the asset-driven logistics to service-driven logistics. Under such context, we build a Newsvendor-based inventory model with partial pooling, induced by responsiveness requirements, followed by a pragmatic heuristic inventory solution, W-solution. Through numerical experiments we investigate the advantage of the decentralized network and W-solution over the centralized network and an allocation-based inventory model, a pre-allocation model. The results demonstrate that decentralization can increase profit significantly by capturing more demand through more responsive fulfillment. That is, extra sales exceed the cost saving from risk-pooling in centralized system. The W-solution saves extra cost by exploiting pooling more smartly than the pre-allocation model, especially under decentralized network. We also show that the W-solution does not always result in less inventory than the pre-allocation model but rather it depends on sales margin unlike the common rule of thumb for inventory pooling. Furthermore, through the case of a drop-ship manufacturer in the US market, we provide empirical evidence of the benefit of network decentralization over the risk pooling when responsive fulfillment is required.