Double Fairness Policy Learning: Integrating Action Fairness and Outcome Fairness in Decision-making
Zeyu Bian, Lan Wang, Chengchun Shi, Zhengling Qi
TL;DR
The paper introduces Double Fairness Learning (DFL), a policy-learning framework that simultaneously targets action fairness, outcome fairness, and value maximization. By formulating a multi-objective optimization and employing a lexicographic weighted Tchebyshev scalarization, it identifies Pareto-efficient policies even in non-convex settings, with rigorous regret bounds. The work provides necessary and sufficient conditions for the existence of outcome/double fairness policies, specialized guarantees for Equal Opportunity and Counterfactual Fairness, and demonstrates substantial fairness gains with minimal value loss on both synthetic simulations and real datasets (motor insurance and entrepreneurship training). The approach is flexible across fairness notions and supports auditing, enhancing trustworthy decision-making in high-stakes domains.
Abstract
Fairness is a central pillar of trustworthy machine learning, especially in domains where accuracy- or profit-driven optimization is insufficient. While most fairness research focuses on supervised learning, fairness in policy learning remains less explored. Because policy learning is interventional, it induces two distinct fairness targets: action fairness (equitable action assignments) and outcome fairness (equitable downstream consequences). Crucially, equalizing actions does not generally equalize outcomes when groups face different constraints or respond differently to the same action. We propose a novel double fairness learning (DFL) framework that explicitly manages the trade-off among three objectives: action fairness, outcome fairness, and value maximization. We integrate fairness directly into a multi-objective optimization problem for policy learning and employ a lexicographic weighted Tchebyshev method that recovers Pareto solutions beyond convex settings, with theoretical guarantees on the regret bounds. Our framework is flexible and accommodates various commonly used fairness notions. Extensive simulations demonstrate improved performance relative to competing methods. In applications to a motor third-party liability insurance dataset and an entrepreneurship training dataset, DFL substantially improves both action and outcome fairness while incurring only a modest reduction in overall value.