On-Demand Service Sharing with Collective Dynamic Pricing

This paper designs a dynamic allocation and pricing mechanism for an on-demand platform that can provide individual or shared services. Time-sensitive customers arrive stochastically onto the platform with heterogenous willingness to pay. The mechanism optimizes at what time, to which customers and at what price to provide services. This environment creates inter-agent and inter-temporal dependencies. We decompose the platform's problem into a dynamic program, based on the novel notion of collective virtual value --- defined as the total surplus that the platform can extract from all customers. The optimal mechanism in this complex environment follows a simple, easily-implementable index rule: service is provided each time the collective virtual value exceeds a threshold, which decreases with the number of available suppliers. Service sharing thus leads to temporal discrimination: as opposed to the classical problem of dynamic pricing with commitment, customers may receive an immediate service or a delayed service --- based on their own willingness to pay and the system's dynamics. Ultimately, service sharing is not exclusively governed by cost-minimization and demand-supply management objectives, but also by discriminatory incentives: it provides an additional degree of freedom to offer differentiated service levels across heterogenous customers