Management of product variety in retail operations

In the first part of the dissertation, we study the assortment planning problem where the retailer selects the set of products to be carried at each store and the inventory level of each selected product. We introduce a novel methodology of demand estimation for multiple products under demand substitution using sales data from multiple stores. We obtain a chain-wide substitution rate and for each store, we obtain demands of products carried in the store as well as products previously not carried in the store. We also describe a heuristic optimization procedure for localizing the assortment to each store and establish new structural properties (based on the heuristic solution). In the second part, we study the interaction of variety decisions across different merchandise categories and their impact on the long run attractiveness of the stores to the customers. We characterize the equilibrium and optimal variety levels under decentralized and centralized management regimes, respectively. We show that category management, which is an example of a decentralized regime for controlling assortment, chooses variety levels that are lower than the optimal solution in at least one category and the profit loss due to decentralization is significant. We describe a simple heuristic that recovers most of this loss. In the third part of the dissertation, we explore the impact of competition between suppliers selling their products through a retailer under wholesale price contracts and revenue sharing contracts. Our model with competing contracts brings new perspectives to the area of supply chain coordination with contracts. Contracts serve as a tool for competition rather than achieving system optimality. Under wholesale price contracts, we show that introducing competition at the suppliers can be desirable as it offsets some of the double marginalization effects. We find that the revenue sharing contracts coordinate the system when the retailers are centrally managed, but lead to more than system optimal inventory when retailers are decentralized. Also revenue sharing contracts lead to more inventory than wholesale price contracts