Compliance Moral Hazard and the Backfiring Mandate

arXiv:2604.21789v1 Announce Type: cross Abstract: Competing firms that serve shared customer populations face a fundamental information aggregation problem: each firm holds fragmented signals about risky customers, but individual incentives impede efficient collective detection. We develop a mechanism design framework for decentralized risk analytics, grounded in anti-money laundering in banking networks. Three strategic frictions distinguish our setting: compliance moral hazard, adversarial adaptation, and information destruction through intervention. A temporal value assignment (TVA) mechanism, which credits institutions using a strictly proper scoring rule on discounted verified outcomes, implements truthful reporting as a Bayes--Nash equilibrium (uniquely optimal at each edge) in large federations. Embedding TVA in a banking competition model, we show competitive pressure amplifies compliance moral hazard and poorly designed mandates can reduce welfare below autarky, a ``backfiring'' result with direct policy implications. In simulation using a synthetic AML benchmark, TVA achieves substantially higher welfare than autarky or mandated sharing without incentive design.

Source: arXiv cs.LG | 2026-04-24