发表机构
Osaka Metropolitan University; TUMSAT（大阪公立大学; 东京都市大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文研究带有原始与调整优先级顺序的择校问题，提出弱稳定概念并利用效率调整延迟接受算法，得到无法被帕累托改进的弱稳定匹配，且机制对更强调整具有响应性。
AI 中文摘要
本文研究了当每所学校既有一个原始优先级顺序，又有一个反映平权行动等政策的调整后优先级顺序时的择校问题。我们引入了一个弱稳定性的概念，该概念允许在一个优先级顺序下存在合理嫉妒，但前提是相反的优先级关系得到另一个优先级顺序的支持。我们将弱稳定匹配刻画为在两个优先级配置的交集下的稳定匹配。利用效率调整的延迟接受算法，我们获得了一个弱稳定匹配，该匹配在弱稳定匹配集合内，对于任何包含所有因调整而受益学生的群体，都无法被帕累托改进。我们还表明，任何对该结果的帕累托改进都必然会在两个优先级配置下产生合理嫉妒。最后，我们的机制对更强的优先级调整具有响应性：在更强的平权行动政策下的结果，对于任何包含所有因更强调整而受益学生的群体，不会被较弱政策下的结果所帕累托支配。
英文摘要
This paper studies school choice when each school has an original priority order and an adjusted priority order reflecting policies such as affirmative action. We introduce a weak notion of stability that permits justified envy under one priority order only when the opposite priority relation is supported by the other. We characterize weakly stable matchings as stable matchings under the intersection of the two priority profiles. Using the efficiency-adjusted deferred acceptance algorithm, we obtain a weakly stable matching that cannot be Pareto improved upon, within the set of weakly stable matchings, for any group containing all improved students by the adjustment. We also show that any Pareto improvement over this outcome necessarily creates justified envy under both priority profiles. Finally, our mechanism is responsive to stronger priority adjustments: the outcome under a stronger affirmative action policy is not Pareto dominated by that under a weaker policy for any group containing all students improved by the stronger adjustment.
