Do Electoral Cycles in Police Hiring Really Help Us Estimate the Effect of Police on Crime ?

This paper replicates a well-known article by Steven Levitt (AER, 1997). Levitt argues that an electoral cycle in police hiring generates exogenous variation in city police force size which may be used to estimate the causal effect of police on crime. I point out that a weighting error in Levitt’s computer program led to incorrect inferences for the key results of his paper. Replication estimates using the correct (and intended) weighting scheme are all statistically insignificant at even the 10% level. In addition to correcting the weighting error, I present results based on a new mayoral election year indicator. This new measure suggests that the electoral cycle in police hiring is (moderately) stronger than reported in Levitt. However, twostage least squares estimates using this new measure are even less precise than (correctly weighted) estimates using Levitt’s original measure, despite the stronger first-stage relationship.