Economics & Econometrics Seminar Series presents
Balanced Distribution Matchings Using Serial Choice Algorithm and Peer-Dependent Capacities
Dr. Edward Honda, University of Manitoba
Friday, March 15, 2024
2:30 p.m. - 4:00 p.m.
307 Tier Building
Theoretical work on many-to-one matchings has a wide range of practical applications. Examples include markets for matching students to public schools or medical residents to hospitals. In many of such applications, we expect there to be distributional constraints to avoid some institutions (hospitals) hiring a large number of individuals (doctors) while others do not have enough. We introduce a class of distributional constraints in which the capacity of a hospital could depend on the number of doctors hired by other hospitals. Although the class of constraints is too general to guarantee a feasible matching that is individually rational, fair, and non-wasteful, we develop an algorithm called the Serial Choice Algorithm to show that a feasible matching with the desired properties exists if we slightly weaken non-wastefulness. In addition to generalizing standard fixed capacities, our class of constraints can model settings such as regional capacities. It can also ensure that floor constraints, and more generally, regional floors are satisfied when we specify the constraints properly and use the Serial Choice Algorithm. We further discuss properties of our model and algorithm including manipulability.