Subgroup fairness in two-sided markets.

Area of Science:

• Economics
• Computer Science
• Operations Research

Background:

• Two-sided markets often exhibit unfairness, with subgroups like female drivers earning less than male counterparts.
• Existing market mechanisms do not adequately address pay disparities among different demographic groups.

Purpose of the Study:

• To propose a novel market-clearing mechanism for two-sided markets that promotes pay equalization across subgroups.
• To introduce and integrate novel fairness concepts: Inter-fairness (subgroup fairness) and Intra-fairness (within-subgroup fairness).
• To optimize market clearing by incorporating customer utility (Customer-Care) alongside fairness objectives.

Main Methods:

• Developed a market-clearing objective function incorporating Inter-fairness, Intra-fairness, and Customer-Care.
• Addressed the non-convexity of the objective function using a non-convex augmented Lagrangian relaxation.
• Leveraged semidefinite programming and the concept of "hidden convexity" to approximate the solution efficiently.

Main Results:

• Demonstrated that the proposed mechanism can be implemented efficiently with a polynomial time complexity relative to the number of participants.
• Successfully applied the mechanism to a simulated ride-sharing system (Uber-like) for driver-ride assignment.
• Showcased the efficacy, scalability, and trade-offs between Inter-fairness and Intra-fairness in practical scenarios.

Conclusions:

• The novel market-clearing mechanism effectively promotes pay equalization across diverse subgroups in two-sided markets.
• The approach offers a scalable and efficient solution for complex market optimization problems with fairness constraints.
• The framework allows for balancing different fairness notions and customer utility, paving the way for more equitable market platforms.