We examine whether concentrating enforcement resources, thereby increasing sanction probability in targeted areas or times (at the expense of others), effectively reduces socially unwanted activities. Our theoretical model and experiment investigate whether splitting the probability of sanction p into higher p_H and lower p_L can minimize overall unwanted behavior. This study empirically
tests the concept of beneficial splitting, central to Bayesian persuasion literature,
to determine if it offers practical advantages in a realistic, parametrized setting.