Optimal Replenishment in Vendor Managed Inventory Systems Using Point-of-Sales Data

Abstract

Events such as spoilage, expiration, employee theft, and customer shoplifting reduce available inventories in retail stores without these reductions being reflected in inventory records. As a result, inventory records often include phantom inventories, i.e., units of good not actually available for sale. These phantom inventories can cause replenishment delays which lead to stockouts and ultimately hurt service levels. In this paper, we study how inventory managers can fully utilize point-of-sales (POS) data for the design of replenishment strategies that account for the existence of such phantom inventories. The optimal replenishment policy in the presence of phantom inventories is complex. However, analyzing the structure of the optimal replenishment problem we are able to propose a simple policy which rarely acts sub-optimally and hence performs close to the optimal policy. This simple policy is based on a threshold on the estimated fraction of demand to be met on a given day conditional on the POS data up to that day, a statistic that we refer to as the daily expected service level.