Using data collected from geographic locators placed in all police vehicles in Dallas, Texas over a one year period, the author estimates the effect of policing on accident outcomes. The author models the occurrence of an accident as a non-homogeneous Poisson process and differentiates between the immediate effect of police presence and the long-term effect of policing on expectations of future police presence. Estimates suggest that at least two days of high intensity stationary police presence at a given time interval and location can reduce that area's accident rate by almost 40 percent during the following week. It is also found that the presence of a stationary police vehicle can immediately reduce the accident rate by at least 9 percent, while the presence of moving police vehicles can produce the opposite effect. (Author/publisher)
Abstract