Solving Practical Dynamic Pricing Problems with Limited Demand Information

Abstract

Dynamic pricing problems have received considerable attention in the operations management literature in the last two decades. Most of the work has focused on structural results and managerial insights using stylized models without considering business rules and issues commonly encountered in practice. While these models do provide general, high-level guidelines for managers in practice, they may not be able to generate satisfactory solutions to practical problems in which business norms and constraints have to be incorporated. In addition, most of the existing models assume full knowledge about the underlying demand distribution. However, demand information can be very limited for many products in practice, particularly, for products with short life-cycles (e.g., fashion products). In this dissertation, we focus on dynamic pricing models that involve selling a fixed amount of initial inventory over a fixed time horizon without inventory replenishment. This class of dynamic pricing models have a wide application in a variety of industries. Within this class, we study two specific dynamic pricing problems with commonly-encountered business rules and issues where there is limited demand information. Our objective is to develop satisfactory solution approaches for solving practically sized problems and derive managerial insights. This dissertation consists of three parts. We first present a survey of existing pricing models that involve one or multiple sellers selling one or multiple products, each with a given initial inventory, over a fixed time horizon without inventory replenishment. This particular class of dynamic pricing problems have received substantial attention in the operations management literature in recent years. We classify existing models into several different classes, present a detailed review on the problems in each class, and identify possible directions for future research. We then study a markdown pricing problem that involves a single product and multiple stores. Joint inventory allocation and pricing decisions have to be made over time subject to a set of business rules. We discretize the demand distribution and employ a scenario tree to model demand correlation across time periods and among the stores. The problem is formulated as a MIP and a Lagrangian relaxation approach is proposed to solve it. Extensive numerical experiments demonstrate that the solution approach is capable of generating close-to-optimal solutions in a short computational time. Finally, we study a general dynamic pricing problem for a single store that involves two substitutable products. We consider both the price-driven substitution and inventory-driven substitution of the two products, and investigate their impacts on the optimal pricing decisions. We assume that little demand information is known and propose a robust optimization model to formulate the problem. We develop a dynamic programming solution approach. Due to the complexity of the DP formulation, a fully polynomial time approximation scheme is developed that guarantees a proven near optimal solution in a manageable computational time for practically sized problems. A variety of managerial insights are discussed.