国内双重差分法的研究现状与潜在问题

Research Status and Potential Problems of Differences-in-differences Method in China

在国内政策效果评估领域,基于自然实验的双重差分法是应用最广的研究工具,并在近年呈现加速发展的趋势。但通过归纳双重差分法的基本应用条件与国内研究现状,本文发现部分研究出现了内生性、控制组受影响、样本异质性等问题。其中尤以内生性问题较为严重,甚至使部分研究产生了一定的缺陷,所谓的自然实验往往难以成立。本文根据实验理论与相关理论基础,对国内主要研究成果进行分类与评述,并针对所发现的问题提出改良办法。

Differences-in-differences method （DID） based on the natural experiment is the most widely used research tool in the academic literature about effect of policy, and the trend of domestic literatures using DID method has been increasing rapidly in recent years. However, through concluding the basic application conditions and domestic research actuality of DID method, this paper finds that domestic research literatures are intermingled with the good and bad, and there are some serious problems, such as the endogeneity problem, control group related and sample heterogeneity. Particularly, the problem of endogeneity is more serious, even making the academic research shows some certain defect, in which the hypothesis of natural experiment is often difficult to establish. According to related experimental research and theoretical study abroad, we classify and comment the major domestic research literatures using DID method, and raise improvement measures for the potential problems identified.