Spatial Information Sharing on On-Demand Service Platforms: A Behavioral Examination (with Basak Kalkanci)
Abstract: On-demand service (eg. ridesharing or food delivery) platforms communicate information about the demand and supply conditions at different locations in a city with drivers via maps. This information can help drivers identify locations where there is high demand but not enough drivers. Using this information, a driver located in an excess-supply location can decide whether to stay there, or to relocate to an excess-demand location. We study the platform's choice of format to share this information. We consider three formats and examine how they affect the level of demand that the platform can satisfy across the locations as a result of drivers' relocation decisions, using a game-theoretic model and lab experiments. We compare surge information sharing (where the platform only reveals the demand-supply conditions in an excess-demand location to all drivers) with full (where the platform reveals the demand-supply conditions in all locations to all drivers) and local information sharing (where the platform reveals the demand-supply conditions in an excess-demand location only to drivers nearby). Experimentally, we find that with full or local information sharing, the platform achieves comparable matching efficiency to surge despite being dominated in theory.
Background Readings (in no particular order and not exhaustive):