Bio:
Irene Lo is an assistant professor in the department of Management Science & Engineering at Stanford University. She conducts research at the intersection of algorithms and economic theory to design matching markets and assignment processes, with a focus on improving efficiency and equity in non-profit applications. She leads the Stanford Impact Lab on Equitable Access to Education, was one of the program chairs for the 1st ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization (EAAMO '21), and was a co-organizer of the Mechanism Design for Social Good (MD4SG) research initiative.
Abstract:
Motivated by high rates of opt-out in a public school district, we consider the problem of yield management when assigning items to agents with priorities at each item. The goal is to determine how much to overbook or 'inflate' capacities for each item to minimize reassignment costs due to opt-out. Such yield management problems arise in several applications, such as assigning students to schools or colleges, allocating airplane seats, or scheduling medical appointments. We show that the presence of priorities generates a bullwhip effect, where less popular items face higher yield uncertainty due to externalities created by more popular items. We provide a Greedy heuristic for inflating capacities that minimizes total cost, and show that under the resulting capacities the bullwhip effect results in higher rates of overbooking and higher costs for less popular items. Simulations using data from a school district validate our theoretical results, and informed overbooking in the district for the 2023-24 school year.
