Researchers are frequently interested in the causal effect of a treatment that is (quasi-)randomly rolled out to different units at different points in time. This paper studies how to efficiently estimate a variety of causal parameters in a Neymanian-randomization based framework of random treatment timing. We solve for the most efficient estimator in a class of estimators that nests two-way fixed effects models as well as several popular generalized difference-in-differences methods. The efficient estimator is not feasible in practice because it requires knowledge of the optimal weights to be placed on pre-treatment outcomes. However, the optimal weights can be estimated from the data, and in large datasets the plug-in estimator that uses the estimated weights has similar properties to the "oracle" efficient estimator. We illustrate the performance of the plug-in efficient estimator in simulations and in an application to Wood et al. (2020a,b)'s study of the staggered rollout of a procedural justice training program for police officers. We find that confidence intervals based on the plug-in efficient estimator have good coverage and can be as much as five times shorter than confidence intervals based on existing methods. As an empirical contribution of independent interest, our application provides the most precise estimates to date on the effectiveness of procedural justice training programs for police officers.
翻译:研究人员往往对在不同时间点向不同单位随意推广的治疗(qasi-)的因果关系感兴趣。本文研究如何有效估计基于随机治疗时间的内曼尼随机框架的各种因果参数。我们解决了在一组测算器中最有效的估测器,该测算器将双向固定效果模型以及若干流行的普遍差异差异方法嵌入巢中。高效的估测器在实践中不可行,因为它需要了解在预处理结果上的最佳重量。然而,最佳加权数可以从数据中估算出来,在大型数据中设置插座估计器,使用估计重量的估测器与高效估测器具有相似的特性。我们展示了在模拟和对Wood等人(2020年a,b)应用中高效的测算器的性能。高效的测算器在对警官程序司法培训方案进行错开展的研究中是行不通的。我们发现,使用估计标定重量的估测器的中间间隔期比标准有效度要短得多,因为根据顶端程序对警官的测算器的准确度,因此,我们现有的测算期可以提供最短的测算。