Empirical studies in online retail operations and dynamic pricing

This dissertation studies empirical problems in retail operations management and dynamic pricing through three essays. The first essay studies the financial impact of offering faster delivery in online retail. Using econometric policy analysis framework, we study a quasi-experimental setting in which a group of U.S. customers for a large apparel retailer experienced a reduction in delivery time due to the opening of a new distribution center (DC). We show that faster delivery increased sales growth by 0.58% per week following the opening of the new DC, with the effect varying inversely with respect to distance from the new DC. The second essay studies the design of free shipping threshold policy in online retail using transaction data from a major online apparel retailer. We develop models of customer demand and product return behavior that are consistent with empirical data to determine the optimal level of free shipping threshold. In particular, we incorporate a behavior called order padding, in which customers deliberately inflate their orders to qualify for free shipping, and its effect on product return. We analyze the model to show that a free shipping threshold policy is most effective when the retailer faces high product margin, low shipping revenue, low product return probability, and when order padding does not cause customers to delay future purchase. The third essay studies practical issues in large-scale multiproduct dynamic pricing. We partner with a Major League Baseball (MLB) franchise to develop a demand model for its single-game tickets. The demand model is then used to evaluate the effectiveness of dynamic pricing policies. The demand model indicates that due to various practical constraints in pricing, the franchise was unable to benefit from the use of dynamic pricing. We address these issues and use simulation to show that revenue improvement of up to 15% can be achieved through the effective use of dynamic pricing. We also show that a properly calibrated fixed pricing policy based on a detailed demand model can achieve similar levels of revenue improvement as the optimal dynamic pricing policy