Stability and Convergence in Matching Processes for Shared Mobility Systems

Abstract

Improvements in mobile information and communication technologies (ICT) and the increasing presence of innovative shared mobility services such as car-sharing, bike-sharing and ride-sourcing have required transportation modelers to forecast and consider matching processes between agents. The transportation modeling community increasingly views these scenarios as two-sided markets, where mobility service providers (suppliers) endogenously adjust their aspirations and behavior in relation to a dynamic traveler demand response. We adopt this two-sided market perspective in modeling the matching process for ridesourcing matching, where drivers match with riders through a bidding process. Through experiments from an agent-based simulation model, we investigate the convergence and matching stability under varying initial assumptions about system characteristics, such as potential payoff or profit and transaction price. These experiments also show that convergence towards a (relatively) stable state may be self-coordinating but is impacted by the total size of the market. Additionally, with respect to the optimal core matching solution and the nucleolus payoff allocations, systems have more difficulty reaching these states with increasing market size, across wider potential profits and across different transaction price mechanisms. Interestingly, by controlling and minimizing the range of the final transaction price, we can minimize the deviation from the nucleolus profit allocation. Designing and managing shared mobility systems must consider these conditions.