匹配法综述：方法与应用  

Survey on Matching Method: Theory and Applications

作　　者：温兴祥 杜在超[1] 

机构地区：[1]西南财经大学经济与管理研究院

出　　处：《统计研究》2015年第4期104-112,共9页Statistical Research

摘　　要：如何运用观测数据进行因果推断是大部分经济学实证研究需要考虑的问题。匹配法(Matching method)基于实验的思想,通过构建平衡样本来推断变量之间的因果联系,正逐渐得到发展和应用。本文对匹配法的产生背景、基本思想、发展历史和实施步骤做一完整介绍,然后进一步给出我国学者运用匹配法的实例。随着匹配法在理论发展上日趋丰富和完善,其作为经济学实证研究中强有力的因果推断工具理应为我国更多的实证研究者所运用。本综述希望有利于这一运用的实现。For most empirical economics study, it is the foremost to consider to do causal inference based on observational data. Matching method achieves this objective through constructing balanced samples, and it has gradually been developed and applied. This paper introduces the topics about matching method, such as background, basic idea, development history and practice issues. Applications examples used by domestic scholars are illustrated. Due to the improvement in theory, as the matching method has become a powerful causal inference tool in empirical economics, and deserved more attention.

关 键 词：因果推断 处置效应 匹配法 倾向分匹配 

分 类 号：O212[理学—概率论与数理统计]