Do 'Broken Windows' Matter? Identifying Dynamic Spillovers in Criminal Behavior

The “Broken Windows†theory of crime prescribes “zero-tolerance†law enforcement policies that disproportionately target light crimes with the understanding that this will lead to future reductions of more severe crimes. We provide evidence against the effectiveness of such policies using a novel database from Dallas. Our identification strategy explores detailed geographic and temporal variation to isolate the causal behavioral effect of prior crimes on future crimes and is robust to a variety of sources of potential endogeneity. We also estimate the effectiveness of alternative targeting policies to discuss the efficiency of “Broken Windows†inspired policies.