Reserve systems are used to accommodate multiple essential or underrepresented groups in allocating indivisible scarce resources by creating categories that prioritize their respective beneficiaries. Some applications include the optimal allocation of vaccines, or assignment of minority students to elite colleges in India. An allocation is called smart if it optimizes the number of units distributed. Previous literature mostly assumed baseline priorities, which impose significant interdependencies between the priority ordering of different categories. It also assumes either everybody is eligible for receiving a unit from any category, or only the beneficiaries are eligible. The comprehensive Threshold Model we propose allows independent priority orderings among categories and arbitrary beneficiary and eligibility thresholds, enabling policymakers to avoid comparing incomparables in affirmative action systems. We present a new smart reserve system that optimizes two objectives simultaneously to allocate scarce resources. Our Smart Pipeline Matching Mechanism achieves all desirable properties in the most general domain possible. Our results apply to any resource allocation market, but we focus our attention on the vaccine allocation problem.

翻译：预留系统通过设立优先考虑特定受益人的类别，在分配不可分割的稀缺资源时容纳多个关键或代表性不足群体。其应用包括疫苗的最优分配，或印度精英院校少数族裔学生的录取安排。若分配方案能最大化资源分配数量，则被称为"智能分配"。现有文献多假设基准优先级，导致不同类别的优先顺序存在显著相互依赖关系，同时默认所有个体均具备任意类别的资源获取资格，或仅允许受益人具备资格。我们提出的综合阈值模型支持类别间独立的优先顺序及任意受益人/资格阈值，使政策制定者得以在平权行动系统中避免不可比指标的强行比较。本研究提出一种同时优化双重目标的新型智能预留系统——智能流水线匹配机制（Smart Pipeline Matching Mechanism），可在最广泛的领域内实现所有理想性质。该成果适用于所有资源分配市场，本文重点探讨其在疫苗分配问题中的应用。