公共政策或项目的因果效应评估方法及其应用
The Causality Evaluation Method on Public Policies and Programs and Its Application

摘要 基于因果推断模型的政策评估领域越来越受到国内外学者的重视，一场经验研究的可信性革命正在应运而生。本文对因果效应的概念进行详尽剖析，回顾了因果推断所应遵循的主要假设条件，分析了遗漏变量偏误这一关键技术问题对于因果推断所造成的影响及其可能的解决方法。在此基础上，本文结合国内外最新研究进展，阐释了工具变量法、倍差法、匹配法、断点回归设计等计量方法的基本原理，以及相应方法在政策或项目的因果效应评估中的实际应用。在未来，通过科学严谨的设计来识别政策或项目执行的效果仍是社会科学研究者的重要使命，因果推断方法研究也将不断取得新的突破，为政策或项目评估的科学化提供更加严谨、可行的方法技术。
Abstract：Policy evaluation based on causal inference models has been paid increasing attention by scholars from home and abroad， a revolution of credibility in empirical research is taking place. This article explores the concept of causal effect in detail， reviews the main assumptions that causal inference should be complied with， and analyzes the impacts that the key technical problem of omitted variable bias brings to causal inference and the possible solutions. On this basis， this article borrows from recent development in related empirical research， illustrates the basics of some typical econometric methods such as Instrumental Variables， Difference in Difference， Matching， Discontinuity Regression， and their applications in policy and program evaluations. In the future， it would be an important mission for social scientists to identify policy or program effects by scientific research design， the research on causal inference would also be broken through， so as to provide policy and program evaluation with more rigid and feasible references.