论文标题
带有凸壳限制的合成控制方法：贝叶斯最大后验方法
Synthetic control method with convex hull restrictions: A Bayesian maximum a posteriori approach
论文作者
论文摘要
通过观察数据，合成控制方法已在因果研究中越来越流行，尤其是在估计对少数大型单位实施的干预措施的影响时。实施合成控制方法面临两个主要挑战：a）估算每个控制单元创建合成控制和b）提供统计推断的权重。为了克服这些挑战，我们提出了一个贝叶斯框架，该框架以可相当的可移位凸壳实现合成控制方法，并提供了有用的贝叶斯推断，这是从惩罚的最小二乘法与贝叶斯最大值的后质（MAP）方法之间的双重性中得出的。模拟结果表明，与替代方案相比，所提出的方法导致偏差较小。我们将贝叶斯方法应用于Abadie和Gardeazabal（2003）的真实数据示例，并发现治疗效果在治疗后的子集中具有统计学意义。
Synthetic control methods have gained popularity among causal studies with observational data, particularly when estimating the impacts of the interventions that are implemented to a small number of large units. Implementing the synthetic control methods faces two major challenges: a) estimating weights for each control unit to create a synthetic control and b) providing statistical inferences. To overcome these challenges, we propose a Bayesian framework that implements the synthetic control method with the parallelly shiftable convex hull and provides a useful Bayesian inference, which is drawn from the duality between a penalized least squares method and a Bayesian Maximum A Posteriori (MAP) approach. Simulation results indicate that the proposed method leads to smaller biases compared to alternatives. We apply our Bayesian method to the real data example of Abadie and Gardeazabal (2003) and find that the treatment effects are statistically significant during the subset of the post-treatment period.
