I examine the effect of local police militarisation on violent crime using evidence from the 1033-programme in the US. Exogenous cost shifters due to the particular logistics of the programme are exploited to instrument for the amount of equipment received by local law enforcement. The results do not support previous county-level studies, who find strong and consistent negative effects on crime. I show that those findings are likely based on a combination of (i) inconsistencies in the underlying data and (ii) limited comparability of different subsamples. Accounting for these factors, I find only weak evidence of a negative impact on violent crime – notably for more rural areas, which form a majority of US counties. For this subsample, the results do not support the notion that military equipment enhances the effectiveness of enforcement agencies: if anything, arrests fall while any resulting crime reduction is of negligible economic significance.
