The Economist:
Roland Fryer on better alternatives to defunding the police
10 July 2020
Using police presence to estimate the impact of police on crime is exceedingly difficult, but has been done in at least one impressive case study.
Using data from the 2005 London terror attacks, Mirko Draca, Stephen Machin and Robert Witt estimated the causal impact of police presence on crime: a 10% increase in total police hours reduces crime by around 3-4%, and when police deployments returned to their pre-attack levels six weeks later, the crime rate rapidly returned to its pre-attack level.