Crying about a Strategic Wolf: A Theory of Crime and Warning

26 Pages Posted: 4 Nov 2016 Last revised: 10 Jun 2020

See all articles by Aaron Kolb

Aaron Kolb

Indiana University - Kelley School of Business - Department of Business Economics & Public Policy

Vincent Conitzer

Duke University

Date Written: June 8, 2020

Abstract

We analyze cheap talk warnings about a strategic adversary, with applications to cybersecurity and national security. Each period an expert receives a noisy private signal about whether an attack by the adversary is feasible. The expert wants to warn a decision maker while also maintaining credibility for future warnings, but unlike in a standard cheap talk game, the adversary can undermine the expert's credibility by delaying attack. While such delays increase "warning fatigue," they also make the expert less tempted to exaggerate so as to avoid too many false alarms. We show that the net effect of a strategic adversary can be better incentive alignment between the expert and decision maker that benefits them both. Moreover, we show that sometimes the expert and decision maker benefit from the expert's ability to exaggerate, as this can induce more defensive action and more strategic delay.

Keywords: Warning systems, cybersecurity, national security, cheap talk, repeated games

JEL Classification: C73, D83, F52, K42

Suggested Citation

Kolb, Aaron and Conitzer, Vincent, Crying about a Strategic Wolf: A Theory of Crime and Warning (June 8, 2020). Kelley School of Business Research Paper No. 16-69, Available at SSRN: https://ssrn.com/abstract=2863907 or http://dx.doi.org/10.2139/ssrn.2863907

Aaron Kolb (Contact Author)

Indiana University - Kelley School of Business - Department of Business Economics & Public Policy ( email )

Bloomington, IN 47405
United States

HOME PAGE: http://https://sites.google.com/site/aaronmkolb/

Vincent Conitzer

Duke University ( email )

100 Fuqua Drive
Durham, NC 27708-0204
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
75
Abstract Views
804
Rank
569,600
PlumX Metrics