How (Not) to Eat a Hot Potato

Abstract

We consider a dynamic matching problem where players are repeatedly assignedtasks and can choose whether to accept or reject them. Players prefer to avoidcertain tasks (“hot potatoes”) while other tasks give a positive payoff (“sweet pota-toes”). There are frictions in the matching process in that players may not bematched to desirable tasks even if one is available. Both under the optimal mecha-nism and in decentralized settings, players may accept hot potatoes if this reducesfrictions in the matching process. We quantify the welfare loss due to matchingfrictions and show that, unlike losses due to more conventional frictions, it does notvanish when the cost of handling hot potatoes grows large.