Spies in a barrel: When to reel in espionage

How do secrets affect international order? We present a formal model of counterintelligence as policy. In our model, the state can learn foreign agent activities from choices in preceding periods. Agents can moderate these actions to suppress the likelihood of discovery. States will only intervene when espionage exceeds a tacitly-agreed threshold, and excesses emerge when agents lack incentive to moderate activity. Non-intervention or escalation depends on executive capacity to detect, future benefits of positive state relations, and restraint by the intelligence community. Egregious punishment of spies and blowback from an executive’s audience make avoiding escalation harder, and intelligence leaks produce ambiguous effects on the stability of relations. We discuss these findings in the context of historical and popular accounts of covert activities revealed to the public.