Abstract
This paper develops a novel multi-agent reinforcement learning (MARL)framework for reinsurance treaty bidding, addressing long-standinginefficiencies in traditional broker-mediated placement processes. We pose thecore research question: Can autonomous, learning-based bidding systems improverisk transfer efficiency and outperform conventional pricing approaches inreinsurance markets? In our model, each reinsurer is represented by an adaptive agent thatiteratively refines its bidding strategy within a competitive, partiallyobservable environment. The simulation explicitly incorporates institutionalfrictions including broker intermediation, incumbent advantages, last-lookprivileges, and asymmetric access to underwriting information. Empirical analysis demonstrates that MARL agents achieve up to 15% higherunderwriting profit, 20% lower tail risk (CVaR), and over 25% improvement inSharpe ratios relative to actuarial and heuristic baselines. Sensitivity testsconfirm robustness across hyperparameter settings, and stress testing revealsstrong resilience under simulated catastrophe shocks and capital constraints. These findings suggest that MARL offers a viable path toward moretransparent, adaptive, and risk-sensitive reinsurance markets. The proposedframework contributes to emerging literature at the intersection of algorithmicmarket design, strategic bidding, and AI-enabled financial decision-making.