发表机构
University of Toronto; IZA; CESifo（多伦多大学; IZA; CESifo）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文研究如何设计契约以同时揭示专家信念和利益，防止误报，并给出最优阈值设计条件。
AI 中文摘要
预测指导决策，但一个私下重视决策的专家可能会误报信念以影响决策。我们研究同时揭示专家关于可验证状态的信念及其在二元行动中利益的契约。当具有不同利益的代理人可能持有相同信念时，精确的仅基于信念的激励相容性迫使行动忽略报告。联合筛选允许响应性，但可能需要实施租金。对于阈值，最小实施租金等于信念依赖截止点的分布加权正曲率。最优阈值设计是租金调整的凹化。我们给出了无限制阈值最优性的充分条件，并在其之外给出了界限和直接最优性证书。
英文摘要
Forecasts guide decisions, but an expert who privately values the decision may misreport beliefs to influence it. We study contracts that jointly screen the expert's belief about a verifiable state and her stake in a binary action. When two distinct stake types can each hold every belief, exact belief-only IC forces the action to ignore the report. Joint screening permits responsiveness but may require implementation rent. For thresholds, minimum rent equals the distribution-weighted positive curvature of the belief-dependent cutoff. Optimal threshold design is rent-adjusted concavification. We give sufficient conditions for unrestricted threshold optimality and bound possible gains when they fail.
