The Effect of Police on Recorded Crime vs. The Effect of Police on Victimisation of Crime. Evidence for England and Wales

Using two sources of crime data, police statistics on recorded crime and victimization data from the British Crime Survey, we provide evidence that measurement error in recorded crime statistics results in underestimation of the effect of police on violent crime. We do not find a similar estimation bias for the effect of police on property crime. Changes in the way the police record violent incidents rather than changes in reporting behaviour of the public are shown to be the underlying cause of the estimation bias. Our findings provide an explanation for the common finding of studies based on police recorded crime of a negative effect of police on property crime but no effect of police on violent crime. This type of measurement error in police recorded crime has been found in many countries, including the US. To address endogeneity in the relation between police and crime, we model the police funding formula, which is used to distribute police resources across police force areas in England and Wales. We use the difference between actual police levels and police levels predicted by the funding formula to identify the effect of police on crime.