A Markov Decision Process approach for balancing intelligence and interdiction operations in city-level drug trafficking enforcement

Abstract

We study a resource allocation problem in which law enforcement aims to balance intelligence and interdiction decisions to fight against illegal city-level drug trafficking. We propose a Markov Decision Process framework, apply a column generation technique, and develop a heuristic to solve this problem. Our approaches provide insights into how law enforcement should prioritize its actions when there are multiple criminals of different types known to them. We prove that when only one action can be implemented, law enforcement will take action (either target or arrest) on the highest known criminal type to them. Our results demonstrate that: (i) it may be valuable to diversify the action taken on the same criminal type when more than one action can be implemented; (ii) the marginal improvement in terms of the value of the criminals interdicted per unit time by increasing available resources decreases as resource level increases; and (iii) there are losses that arise from not holistically planning the actions of all available resources across distinct operations against drug trafficking networks.