On-demand mobility services have transformed urban mobility. Their optimal management can benefit at individual and collective levels, and their efficient integration within transportation systems could enhance their performance. Conversely, poor fleet management result in high pick-up waiting times, abandonment rates or extra traffic. One challenge for operators is proactively rebalancing their fleets to ensure the supply matches the demand. This paper addresses this issue with an auctioning approach. We design a multi-agent architecture in which local controllers encourage vehicles to relocate nearby. We conduct simulation studies on the city of Lyon in France. Our results evidence a significant increase in the number of passengers served compared to a norebalancing scenario.