We calculated the efficiencies with which Brazilian states provide public safety by using a stochastic frontier cost function and panel data from 2001-2006. We estimate a Translog with fixed effects where prices for the production factors are the initial salaries of military police, civil police, and precinct chiefs and the product is the inverse of the homicide rate. To explain inefficiencies we use economic and demographic variables which measure active drug markets and the allocation of judicial, police, and penitentiary expenditures. Our models indicate that inefficiencies are higher when the size of drug markets, the ratio of military to civil police, the rate of high school dropouts, and income inequality increase. On the other hand, increasing judicial expenditures and the number of prison cells reduce inefficiency. São Paulo is the most cost efficient state providing public safety, whereas the Federal District is the least cost efficient.
Cost-Efficiency; Stochastic Frontier; Public Safety
