Integrated Deterrence: Episodic Analysis
Principal Investigator: Eli Berman, UC San Diego
Co-Investigators: Esteban Klor, Hebrew University, Jerusalem
Years of Award: 2023 - 2026
Managing Service Agency: Air Force Office of Scientific Research
We propose formalizing and empirically testing key aspects of Integrated Deterrence theory, which is central to the National Defense Strategy, and indeed central to conflict in the modern era, ranging from deterring information operations in the grey zone to deterring conventional and nuclear attack.
We will develop the concept of a response function –derived from game theory, allowing us to analyze deterrence by cost imposition by examining the slope of response as a function of damage incurred. We will also formalize and test hypotheses about deterrence by denial and deterrence by resilience, using shifts of that response function. By examining the shapes of interacting response functions we can find equilibria, understand escalation, and inform escalation management. Our approach will integrate both economic and humanitarian tools with destructive activity, broadening conventional deterrence theory to allow “frenemies” –who are simultaneously in conflict, competition and cooperation.
Our empirical analysis is enabled by a unique dataset under development, which records the very high frequency of attacks along the Gaza – Israel border, collected by neutral observers, from 2007 through 2018. Measurement occurs at a frequency of up to five minutes, and includes attacks from multiple IDF platforms and multiple Gazan protagonists, both using a variety of armaments (including missiles, rockets, mortars, airstrikes, incursions, etc…), including casualties and fatalities of both combatants and noncombatants. It spans large Israeli incursions --conducting deterrence by denial, as well as the introduction of “Iron Dome” air defense –providing resilience. Those data are supplemented with data on border crossings, reflecting bilateral trade and humanitarian activity, adding a frenemies dimension. An additional attractive aspect of these data is that we can discuss it with decisionmakers on at least one side and cross-validate it with social media of all protagonists.
Preliminary analysis based on a partial version of those data reveals that the conflict is best understood as a sequence of attacks and counterattacks aggregated to episodes. Episodes have an internal logic of stimulus-response in which calendar time is of secondary importance. That structure necessitates creating new techniques in computer-aided coding to establish sequence, and in statistically analyzing episodic data to deal with serially correlated errors between actions (rather than time periods).
These techniques, piloted on preliminary data, indicate stable response functions over hundreds of episodes, once the intensity of attack is aggregated across munitions to create common damage measures.
Those (preliminary) response functions indicate stable deterrent postures of both sides –Israel and Gaza. Consistent with this posture, Israel disincentivizes violence by opening border crossings during lulls. The conflict has a single, stable equilibrium, though at a positive level of violence –indicating incomplete deterrence. Both sides practice mixed (i.e. probabilistic) strategies, so that their stable equilibrium is not a point but a two-dimensional distribution in a damage space.
Our analytical results –which rely on the shape of response functions, should be of general use in analyzing deterrent interactions, including aspects of deterrence (cost, denial, resilience) and escalation management. We are fairly confident that this project can yield analytically tractable tools to inform Integrated Deterrence theory generally, and lead to practical insights for measuring and analyzing response functions of competitors, and ultimately for guiding de-escalation towards more constructive equilibria.