报告题目：Joint Assortment and Pricing under MMNL: From Full Customization to Fair Pricing
报告人：Queen’s University, Smith School of Business 李光副教授
主持人：秦绪伟教授
报告时间：2026年8月4日（周二）10:00--11:30
报告地点：文管楼B306
报告摘要：
Personalized assortments are common and widely accepted, but charging different customers different prices for the same product is often perceived as unfair and increasingly draws regulatory scrutiny. Motivated by this tension, we study joint assortment and price optimization under the mixture of multinomial logit (MMNL) model across the full spectrum of pricing fairness. Three regimes anchor the analysis: fully customized prices and assortments; fair, common prices with customized assortments; and fully fair prices and assortments, which is the standard MMNL joint problem.
We develop a unified algorithmic framework built on two ideas: a multiplicative discretization of the continuous price space, and a reformulation over virtual products, each a product offered at a discrete price level, that recasts joint pricing and assortment as a constrained assortment problem. For the fully fair MMNL problem, which is difficult because assortment optimization is NP-hard and the pricing objective is not quasiconcave, we obtain a fully polynomial-time approximation scheme that requires no structural assumptions on the price-utility functions, allows a capacity constraint, and extends to the multinomial and nested logit models. For fair prices with customized assortments, we combine the reformulation with a geometric decomposition, partitioning the segment-revenue space into hyper-rectangles and, within each, a hyperplane arrangement into polynomially many cells, so that the problem reduces to a polynomial number of linear programs and is solved to near-optimality in polynomial time for a fixed number of customer types.
Finally, we quantify the cost of fairness. We prove tight bounds showing that each fairness restriction lowers revenue by at most a factor equal to the number of customer segments, a worst case that is achievable, yet in a broad range of instances the loss is only a constant. Fair pricing with customized assortments can therefore capture much of the value of full personalization while avoiding price discrimination.
报告人简介：
Guang Li is Academic Director (Master of Management Analytics), an Associate Professor and Distinguished Research & Teaching Fellow of Management Analytics at Smith School of Business, Queen’s University. Her research interests includechoice modeling, revenue management, assortment planning, retail operations, supply chain management,interfaces between management analytics and other business disciplines.Guang’s research has been funded by the Natural Sciences and Engineering Council of Canada. Her work has been published in leading academic journals, including Operations Research,Management Science, Manufacturing & Service Operations Management, and Production and Operations Management. She is a Senior Member of INFORMS. Guangs has also served as track chair and session chairs at POMS and INFORMS Annual Meetings.
欢迎有兴趣的师生积极参加！
