Scholars have debated whether executive power sharing can secure peace in multiethnic states, but concerns about endogeneity due to reverse causation render this a difficult problem for empirical analyses. In the absence of a suitable instrumental variable, I explore an actor-based approach by studying a simple formal model. This highlights the conditions under which governments are likely to share power with a domestic challenger depending on the threat of violence. I then formulate a statistical “strategic selection” model that closely mirrors the theoretical model, thereby directly incorporating endogeneity. Applied to data at the level of ethnic groups, the estimation results indicate that power sharing is indeed enacted strategically by governments in anticipation of the risk of conflict. However, shedding new light on an ongoing debate, I find that the critics have overstated the case against power sharing: rather than spurring it, power sharing robustly reduces civil conflict.

中文翻译：

---

面对内战的行政权力共享

学者们一直在争论行政权力分享是否可以确保多民族国家的和平，但对反向因果关系引起的内生性的担忧使这成为实证分析的难题。在没有合适的工具变量的情况下，我通过研究一个简单的形式模型来探索基于参与者的方法。这凸显了政府可能根据暴力威胁与国内挑战者分享权力的条件。然后，我制定了一个统计“战略选择”模型，该模型与理论模型非常接近，从而直接结合了内生性。应用到族群层面的数据，估计结果表明，权力分享确实是政府在预测冲突风险时战略性地制定的。然而，为正在进行的辩论提供新的思路，