Public debates over the death penalty in the U.S. have long focused on the punishment's putative deterrent effect. Unsurprisingly, numerous scholars have sought to determine (1) whether the penalty in fact prevents murders, and (2) how many murders it prevents. However, these efforts have deployed a fairly homogeneous set of methodological techniques—primarily OLS and two-stage least squares with instrumental variables. This article breaks from that tradition by using synthetic controls to assess the deterrent capacity of capital punishment. Applying this technique using seven states that recently abolished the death penalty and 29 states that retained the punishment during the same period, I find no evidence that the presence of a capital punishment statute in a state is sufficient to deter murders. These results are robust to numerous alternative specifications; they also persist when I use stranger homicides—which are theoretically more susceptible to deterrence—as the dependent variable.

中文翻译：

---

死刑法规和谋杀率：来自合成控制的证据

美国关于死刑的公开辩论长期以来一直集中在死刑的推定威慑作用上。不出所料，许多学者试图确定（1）刑罚实际上是否可以防止谋杀，以及（2）它可以防止多少起谋杀。然而，这些努力已经部署了一套相当同质的方法技术——主要是 OLS 和带有工具变量的两阶段最小二乘法。本文打破了这一传统，通过使用综合控制来评估死刑的威慑能力。使用最近废除死刑的七个州和同期保留死刑的 29 个州应用这种技术，我发现没有证据表明一个州的死刑法规足以阻止谋杀。这些结果对许多替代规范都是稳健的；当我使用陌生人凶杀案（理论上更容易受到威慑）作为因变量时，它们也会持续存在。