Police & Thieves: geographical crime profiling

It is a well established empirical result on the economics of crime that preventive and punitive measures reduces the incidence of criminal activities (Levitt, 1996; Cook, 2009; Draca and Machin, 2015). In particular, an increase of the size of the police is a possible channel to reduce crime (Levitt, 1997, Evans and Owens, 2007). Two of these works (Di Tella and Shargrodsky, 2004; Klick and Tabarrok, 2005) study the effect of the displacement of police units on the geographical crime profiling. Following the events of terrorist attacks respectively in Buenos Aires and in London, policing in the attacked areas became more intensive. Although crime in these areas dropped after these changes, overall crime in both cities remained constant: the new configuration of police units displaced crimes into regions that became underprotected. The evidence presented in these works illustrates the importance of considering the spatial allocation of police units as a crucial part of the policing strategy. The probability of catching criminals is indeed an object of equilibrium coming from decisions of where to commit crimes and where to place police units. Indeed, it is typical that more than 50% of the reported crimes in bigger cities occur in less then 5% of the streets — the hot-spots of crime (Weisburd 2005; Collazos et al, 2019). Surprisingly enough, most of the papers dealing with this issue are too stylized. Papers dealing with equilibrium environments (Freeman, Grogger, and Sonstelie, 1996; Lazzati and Menichini, 2016; Algahtany, Kumar, and Barclay, 2017) divide cities into regions and the decisions are made at region-level. They all explain crime across regions, but none of them is able to explain what happens inside of the regions. Some other papers aim at explaining the occurrence of crime at local-level without equilibrium considerations (Short et al, 2008; Mohler et al, 2011), i.e., in their papers, agents are not necessarily optimizing when they choose where to commit crimes. A well-designed police allocation strategy has an important property in terms of policy. Changing the geographical configuration of the police does not imply a higher level of expenditures in police. Therefore, any reduction on property and violent crime rates (these last ones particularly costly to the society) under a new distribution of police units is highly cost-effective.

I will produce a policy-oriented paper containing an original framework modelling criminal and police decisions by point processes. The appropriateness of the models will be tested using available data on crime and police locations.