Abstract The synthetic control approach to analyze policy impact is here complemented to compute i) multivalued treatment effects and ii) treatment effect evaluated at different levels of the distribution of the treated, at the various quantiles. Confining the analysis to binary treatment and/or to average values alone can be very misleading since treatments can be multivalued and they can have different impact away from the mean, in the tails. Two case studies and a set of simulations allow to investigate the behavior of the proposed approach.

中文翻译：

---

综合控制方法：分位数的多值处理

摘要 分析政策影响的综合控制方法在这里得到补充，以计算 i) 多值处理效果和 ii) 在不同分位数的被处理分布的不同水平上评估的处理效果。将分析仅限于二元处理和/或单独的平均值可能会非常具有误导性，因为处理可以是多值的，并且它们可能会在尾部产生远离平均值的不同影响。两个案例研究和一组模拟允许调查所提出方法的行为。