Topic: Refund or Not?
Joint Pricing and Refund Optimization for Omnichannel Retailing with Product Returns
Time: 14:30 PM-16:00PM, June 9, 2025
Location: Room A1148, New Main Building
Guest: Dr.Zhenzhen Yan is an associate professor at School of Physical and Mathematical Sciences, with a courtesy appointment at Nanyang Business School, Nanyang Technological University.Her research interests primarily focus on data-driven resource allocation.She has pioneered a series of robust and responsive methods,especially effective with moderate datasets. Her work has broad applications, spanning smart city operations, supply chain management, and e-commerce operations. Through the development of advanced models and analytical tools, she addresses real-world challenges, particularly in decentralized and dynamic environments, ensuring solutions are both scalable and practical. Her work has led to more than 10 publications in leading operations management journals including Management Science, Operations Research,MSOM and POMS. Her research excellence has garnered recognition through numerous accolades, including 2020 MSOM Data-Driven Research Challenge (as a finalist), the 2022 SPMS Young Researcher Award, and the 2023 Asia-Pacific Operational Research Society Young Researcher Best Paper Award. Her work has also been featured in major media outlets like The Straits Times and ScienceDaily. Currently, she serves as an Associate Editor for Decision Sciences. In addition to her research，she is an outstanding educator, receiving the prestigious 2024 Nanyang Education Award (School) for teaching excellence.
Abstract:
We consider an omnichannel retailer selling multiple substitutable products through an online channel and a physical store. Online purchases can be retuned either by mail or to the physical store. The retailer decides each product’s price and refund value for each channel to maximize his expected profit.
We capture a consumer's seguential decisions on making a purchase and potentially returning her product using a generalized Markovian logit choice (MLC) model. We use this model to formulate the retailer's joint pricingand refund optimization problem. lf there are constraints on prices and refund values, then this problem may become non-convex, and we approximate it using a mixed-integer linear program (MlLP). Furthermore, we analytically derive the performance accuracy of MlLP. Otherwise, this problem is convex, and we analytically obtain its optimal solution. We estimate the generalized MLC model using transaction data, making our framework applicable to a data-driven setting. Numerical experiments using synthetic data demonstrate that our estimation-and-optimization framework based on the generalized MLC model fits a general data set well. A case study using data from a major fashion retailer demonstrates that our framework can handle situations with partially observable data and is flexible to incorporate various practical refund policies.
We find that to benefit from return policies, the retailer needs to guarantee good quality or non-negative online (or offine) return social welfare for each product. We also find that products with a smaller loss coefficient result in a higher return rate but a larger profit.
