This study explores the causes of crime and the differences in deterrent effects of policing on crimes among rural, suburban, and urban communities. We hypothesize that certain numbers of all crimes are unaffected by policing due to their high net return; policing deters only marginal crimes. That is, unlike other research efforts, we recognize that there is a level of crime indigenous to a given type of community about which little can be done, although a particular community can affect deviations from this level. By introducing this `natural rate' of crime we are able to empirically reveal the deterrence effect of police expenditures upon all types of property crimes except robberies. The study analyzes 230 communities in a system of six simultaneous equations, using police, crime, and other socio-economic variables. The model can be used by state and regional policy-makers to more effectively allocate resources to the different types of communities under their jurisdiction and among the various police functions designed to deter specific types of crime.