Dashboard Mechanisms for Online Marketplaces

Опубликовано 10 апреля 2018, 19:02

We consider the problem of designing mechanisms for online marketplaces for long-lived agents with persistent preferences. Bidder dashboards, which forecast outcomes as a function of bid, help bidders optimize their bids. The principal in such a scenario can select both the dashboard and the mechanism. We describe a natural behavioral model of bidders in dashboard mechanisms and show that under this model, the principal can implement nearly efficient market outcomes quite generally. Under the classical behavioral model of Nash equilibrium, we show a similar implementation result for single-item and multi-unit environments (where bidders are substitutes with respect to feasibility) and suggest that the same is not possible when bidders are complements with respect to feasibility. Joint work with Aleck Johnsen, Denis Nekipelov, Zihe Wang, and Onno Zoeter.

See more at microsoft.com/en-us/research/v...