Sid Banerjee
The rise and success of online marketplaces, like AirBnB, Lyft and Uber, presents many new and interesting market design problems, going beyond traditional price-based market design. In this talk, I will present two recent projects on these lines:
Discovery-based market segmentation: In many settings, a platform has full control over search and discovery, but prices are endogenously determined by buyers and sellers - for example, AirBnB chooses what listings are shown to a guest, but prices are set by the hosts, and matches determined by guests. Instead of pricing, such platforms can instead use directed-discovery mechanisms (search/recommendation/etc.) to control which buyers/sellers are visible to each other. We consider a model for such discovery-based marketplace segmentation, for which we develop efficient approximation algorithms for maximizing the volume of trade and welfare. This is joint work with Srinivas Gollapudi and Kostas Kollias at Google, and Kamesh Munagala at Duke.
Artificial currency mechanisms: In other platforms, such as those for allocating internal resources (e.g., cluster time in a computing facility/limited seats in a class/etc.) or public goods (vaccines among hospitals/food donations among food banks), the use of money itself is considered inappropriate. We consider the design of repeated allocation mechanisms based on artificial currencies, where we first allot each agent an endowment of credits, which they can use over time to bid for items. We show that a simple mechanism, based on a repeated all-pay auction with personalized endowments, simultaneously guarantees vanishing gains from non-truthful bidding as well as vanishing loss in efficiency.
This is joint work with Artur Gorokh and Kris Iyer at Cornell.