Booster Draft Mechanism for Multi-Object Assignment
We describe a new mechanism - what we call a booster draft - for allocating multiple, indivisibleobjects among a group of individuals. The mechanism’s appeal lies in its strategy-proofness andsimplicity: Individuals take turns drawing objects from different sets - called boosters - and simplyneed to identify their favorite object when it’s their turn to choose. Following a market designapproach, we examine how to tailor the booster draft mechanism to specific multi-object assignmentproblems. As an illustrative example, we consider the assignment of teaching positions to graduatestudents. We show that, through the right design of the boosters, not only is the mechanismstrategy-proof, but the resulting allocations are fair and efficient. In fact, in the described domain,under some additional mild axioms, any strategy-proof mechanism is some variation of a boosterdraft. Finally, using data on graduate students preferences, we demonstrate that the booster draftis useful and easy to implement in practice