基于微观非实验数据的政策效应评估方法评价与比较  

Review and Comparison of Researches on Policy Evaluation Based on Micro Non-experimental Data

作　　者：卫梦星[1] 

机构地区：[1]中国社会科学院研究生院,北京102488

出　　处：《西部论坛》2012年第4期42-49,共8页West Forum

基　　金：中国社会科学院重大课题(YZDA2010)"我国应对金融危机的政策效应分析及退出机制设计";中国社会科学院经济政策与模拟重点研究室项目

摘　　要：社会政策是政府发挥职能的重要手段,对政策实施的效果进行量化评估,有助于政府决策的科学化、合理化。本文基于非实验数据,归纳了目前微观政策领域各种致力于构建"反事实"的政策效应评估方法,包括工具变量法(IV)、断点回归(RD)、双重差分(DID)和匹配方法(Matching),阐述了其各自的优势、劣势及适用条件,并对实际政策评估中各模型的选用进行了简要述评。Social policy is an important means for governments to realize their function. Quantitative evaluation on the effect of policy implementation can help the governments make decisions scientifically and reasonably. Based on non-experimental data, this paper summarizes the evaluation methods for all kinds of the ＂counterfactual＂ policy effects deliberately constructed in current micro-policy field, such as instrumental variable method, regression discontinuity method, difference-in-difference method and matching method, elaborates the advantages, disadvantages and applicable conditions of each method, and makes brief review of the selection of each model in the practice of policy evaluation.

关 键 词：政策效应评估 反事实 准实验方法 匹配方法 工具变量法 断点回归 双重差分 

分 类 号：O212.1[理学—概率论与数理统计] F123.16[理学—数学]