Ezra’s Round Table / Systems Seminar: Manxi Wu (Cornell)
Information and Incentive Design in Congested Networks
Abstract: In this talk, we study the role of information and incentives design in congested networks, aiming to steer the aggregate behavior of strategic agents towards system-wide efficiency and optimal resource utilization. In the first part of the talk, we study how ride-hailing platforms can improve revenue by incentivizing the reposition of drivers via strategically revealing the information of demand shock in networks. In particular, we focus on characterizing practical scenarios, where the optimal outcome is achieved by a simple information mechanism with monotone partitional structure. In the second part of the talk, we study the effectiveness of incentive mechanisms in mitigating traffic congestion. Motivated by the expansion of high occupancy toll lane systems in California, we build a game-theoretic model to study how travelers with heterogeneous value of time and carpool disutilities respond to the dynamic toll prices on the high occupancy toll lanes and the resulting reduction in congestion and emission. We present an approach to estimate the distribution of travelers’ heterogeneous preferences using inverse optimization, and demonstrate the effectiveness of optimal dynamic toll pricing on Highway I-880 using data provided by Caltrans.
Bio: Manxi Wu is an Assistant Professor at Cornell University’s School of Operations Research and Information Engineering. Her area of research includes information and market design, and multi-agent learning in societal networks. Manxi earned her Ph.D. from MIT IDSS in 2021. She was a research fellow at the Simons Program on Learning and Games, and a postdoctoral scholar at EECS, University of California, Berkeley from 2021-2022.