Catastrophe and Rational Policy: Case of National Security

Abstract

Predicting catastrophes sometimes involves heavy-tailed distributions with no mean, eluding proactive policy since expected cost-benefit based analysis fails. We study US government counterterrorism policy, given heightened risk of terrorism. However, terrorism involves human behavior. We synthesize the behavioral and statistical aspects within an adversary-defender game. Calibration to extensive data shows that sometimes Weibull with its heavy tail, but finite mean, fits best and rational policy is feasible. Here, we find US counterterrorism expenditures are nearly optimal and then estimate terrorists’ unobserved parameters, e.g. difficulty to attack. Other times Generalized-Pareto, with no mean, fits best and rational policy fails. Here, we offer practical “work-around”.

Keywords: public policy, catastrophe, national security, terrorism, heavy tail distributions, game

JEL Classification: H56, D81, C46