Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces Conformal Risk Sharing, a novel framework for distributing the financial impact of rare adverse events among participants while ensuring that no individual is made worse off. By formalizing the Certified Allocation Problem, the authors develop a method that combines an interpretable sharing policy with split conformal calibration to provide distribution-free obligation caps for each participant. Experiments demonstrate that this approach effectively mitigates extreme financial burdens for high-risk agents while maintaining fairness across the group.
Participants can now share financial risks from rare events without fear of being worse off, thanks to a new method that guarantees individual obligation caps.
Sharing the financial impact of rare adverse events across a group can soften extreme individual burdens, but any participant made worse off by the arrangement has reason to leave. A credible mechanism must therefore provide each agent with a trustworthy cap on their future obligation and should be deployed only if the aggregate harm across participants is bounded. We formalise this as the Certified Allocation Problem: from finite data and without distributional assumptions, find a redistribution rule, produce obligation caps for every participant, and verify that no participant is made materially worse off. We propose Conformal Risk Sharing, which solves this problem by pairing an interpretable sharing policy with split conformal calibration. The sharing intensity is tuned on training data, while held-out calibration data produces distribution-free per-agent guarantees (valid under exchangeability). Experiments on synthetic and real-world data, including precipitation and energy-cooperative data, confirm that the framework can substantially reduce extreme obligations for high-risk agents while controlling harm to others.