How Bayesian Persuasion can Help Reduce Illegal Parking and Other Socially Undesirable Behavior
We consider the question of how best to allocate enforcement resources across different locations with the goal of deterring unwanted behaviour. We rely on “Bayesian persuasion” to improve deterrence. Our approach is distinguished by the following five features: (1) we consider a problem in which the principal has to allocate resources and then send messages (persuade) rather than just persuade. (2) Messages are received by drivers in n different neighborhoods, so persuasion is with respect to multiple audiences. (3) The problem is a “constrained convexification” rather than just a convexification problem, where the constraints are due to resource and probability restrictions. This implies that convexification may be partial rather than complete as is usually the case in Bayesian persuasion models. (4) Even though the basic problem is not linear, we show that it can be cast as a linear programming problem. Finally, (5) we characterize the number of messages needed in order to obtain the optimal solution, and describe conditions under which it is possible to explicitly solve the problem with only two messages.
(Joint with Penénlope Hernández)