基于约束偏好下的双边匹配理论与实验研究

Stylized two-sided matching models assume that agents can list all potential partners whom they find acceptable. A matching mechanism then matches two disjoint sets of agents with each other based on their full preference lists. However, in practice matching markets typically impede an agent to submit a preference list that contains all his acceptable alternatives. This induces a preference revelation game in which agents need strategically reveal their preference lists with a limited number of acceptable partners. In addition, the acceptance time featured in the family of application-rejection mechanisms may also induce strategic behavior in preference revelation. Based on a typical example of Chinese college admission where students can only apply to a fixed number of colleges, this project aims to provide a detailed study on the performance of the parametric family of application-rejection mechanisms under constrained preferences. On the theory side, we first explore the combined effects of acceptance time and choices quota on preference manipulation incentives. Based on the equilibrium strategic behavior of agents in preference revelation, we subsequently investigate its implications on centralized matching outcomes under different matching mechanisms in terms of efficiency, fairness and social welfare. On the practice side, based on the empirical stylized facts of Chinese college admission system, we test our theoretical predictions using laboratory experiments and further quantify the efficiency loss from constrained preferences. Using a synthesis of game-theoretical and experimental analysis, the project will have important implications for the general constrained mechanism design theory, as well as provide valuable insights for the college admission system reforms in China.

经典双边匹配模型通常假设个体的偏好信息是透明且完备的，市场机制根据偏好信息对其建立相互匹配。而在现实中，个体往往受到约束，只能向市场显露有限多个偏好，如高考志愿填报。在这种约束下，个体需要策略性地显露信息以获得较好的匹配结果。本项研究在经典匹配模型的基础上，探讨在约束偏好下一系列“申请—拒绝”匹配机制（如延迟接受机制、波士顿机制）的综合绩效。首先在理论上研究个体在策略性偏好显露中的博弈行为，而后据此考察并比较不同机制下策略性的偏好显露对匹配结果的效率、公平和社会福利的影响。在实验部分，以高校录取为背景设计一系列匹配实验，模拟不同环境下的“申请—拒绝”机制以检验理论模型的结论，从而进一步量化约束偏好带来的效用损失。通过理论和实验的综合分析，本项研究不仅对解决约束偏好下的一般匹配机制设计问题具有启示作用，而且也将对一系列的实际应用和政策制定（如高校录取、志愿填报机制）提供独特的理论与实证依据。

本项目结合机制设计理论和市场设计应用这两个在微观经济领域中发展最快的分支研究双边匹配和资源分配等重要的社会经济问题。本项研究首先对于不同的匹配机制下偏好显示博弈中的操纵动机、均衡行为以及匹配结果的相关经济性质在理论和实验角度进行评估和比较。随后我们将本项目的研究模型和思路拓展到重要的现实问题，从市场机制设计的角度探讨稀缺资源的分配机制如何影响个体自愿捐献动机的激励政策设计问题。本项研究成果的主要理论贡献在于提出新颖的帕累托稳定匹配算法和机制，并且首次为稀缺资源分配和捐献机制设计建立动态代际交叠模型。基于这些理论框架我们提出一系列优先匹配和分配机制，并分析和比较这些机制对个体决策的激励作用，以及个体决策的变化如何影响匹配结果的公平、效率和社会整体福利。从实践的角度，本项目基于现实匹配数据，通过反事实校准实验和数值模拟实验揭示了允许显示弱偏好可以显著提高匹配效率。本项目重点将理论研究和现实政策需求联系起来，其中所涉及的匹配模型和机制设计模型在现实经济问题中具有广泛的应用，包括大学招生录取、社会融资和器官捐献等市场机制的设计。本项目的研究不仅评估现有匹配机制的作用，而且注重新机制的设计和应用，这对于未来的政策设计将起到积极的建设性作用。