How do states distribute the burdens of collective defense? This paper develops a network theory of burden sharing. We focus on bilateral defense cooperation agreements (DCAs), which promote cooperation in a variety of defense, military, and security issue areas. Using a computational model, we show that DCA partners’ defense spending depends on the network structure of their agreements. In bilateral terms, DCAs increase defense spending by committing states to defense activities and allowing partners to reciprocally punish free riding. However, as a state's local network of defense partnerships grows more densely connected, with many transitive “friend of a friend” relations, DCAs have the countervailing effect of reducing defense spending. The more deeply integrated states are in bilateral defense networks, the less they spend on defense. We distinguish two potential mechanisms behind this effect—one based on efficiency improvements, the other on free riding. An empirical analysis using multilevel inferential network models points more to efficiency than to free riding. Defense networks reduce defense spending, and they do so by allowing countries to produce security more efficiently.

国家如何分配集体防御的负担？本文发展了负担分担的网络理论。我们专注于双边防务合作协议 (DCA)，这些协议促进各种防务、军事和安全问题领域的合作。我们使用计算模型表明，DCA 合作伙伴的国防开支取决于其协议的网络结构。在双边方面，DCA 通过让各州致力于国防活动并允许合作伙伴相互惩罚搭便车来增加国防开支。然而，随着一个州的地方国防伙伴关系网络变得更加紧密，具有许多可传递的“朋友的朋友”关系，DCA 具有减少国防开支的反补贴作用。双边防御网络中一体化程度越高的国家，它们的国防开支就越少。我们区分了这种效应背后的两种潜在机制——一种基于效率提升，另一种基于搭便车。使用多级推理网络模型的实证分析更多地指向效率而不是搭便车。国防网络减少了国防开支，它们通过允许各国更有效地生产安全来实现。