Quantitative models in crime control

In this study we propose several mathematical models for describing the behavior of crime-related systems. In the first part is presented a simple model for describing and measuring crime systems with recidivism. This model may be used in estimating the quantitative effects of alternative policies on crime rates. An important factor not recognized by the model is deterrence. One derivative of the mathematical analyses is a proposed method for measuring the relative severity of sentencing in a given system. The second part of the paper deals with the criminal population as a whole, trying to bring into account the non-stationary behavior of the system. This study should be regarded as a preliminary attempt to develop elementary mathematical models fitting crime systems. Some of the assumptions do not have substantial direct support by data.     

Unitary regulatory supervision or multi-entity supervision? A computational approach to a numbers problem in financial regulation

Policymakers globally have debated (and often implemented) the idea of consolidating numerous financial regulatory supervisory entities into one unitary entity. This article uses a computational model to explore the effect such a decision would have on supervisory performance. Using insights from organizational scholarship on consensus-making among individuals within organizations, the simulation suggests that under most conditions a unitary supervisory entity yields lower performance than smaller, numerous entities with unique mandates, keeping the number of regulatory inspectors constant. This result arises from the heterogeneity of perspectives being shared within the entity and the influence of precedent actions. The results also show a decreasing utility to disaggregation: performance decreases when too few inspectors share among themselves in building consensus. When insufficient heterogeneity within supervisors exists, unitary frameworks outperform multi-entity frameworks. These findings have implications on the design of supervisory frameworks and contribute to research on consensus-building, heterogeneous group membership, and computational modeling.