Stocking Under Dynamic Choice

It is well known that consumers, when faced with limited choices or temporary stock-outs, will often chose different sizes, colors or brands rather than go home empty handed. Intuitively, a firm's decisions about the level of variety they offer and the quantity of inventory they stock ought to account for such behavior. In this talk, we present recent work on how such consumer choice behavior can be incorporated in inventory models and the implications it has for inventory decisions and operating performance.We model customer choice behavior using a very general random utility model. The first problem we consider is that of a monopoly retailer who has to choose inventory levels for a category of products. We analyze structural properties of the profit function and propose a stochastic gradient algorithm for maximizing assortment profits. Our results show that ignoring substitution effects can lead to biases in stocking levels and, in certain cases, significantly reduced profits. We then consider a competitive version of this model in which each variant is managed by an independent firm - for example competing product at a single retail location or the same product sold at competing retail locations. We analyze the Nash equilibrium of the resulting stocking game. We show there is a `` over-stocking" effect caused by competition that leads to zero industry profits as the number of competitors increases. Finally, we use our models to analyze the performance of two supply chain coordination mechanisms - Vendor Managed Inventory (VMI) and Retailer Managed Inventory (RMI) - when there are horizontally competing retailers.