| Summary: | In 2017, over a dozen ferry stations were introduced across the NYC region on multiple dates, serving roughly 10,000 customers per day. We measure a negative association between these stations and crime reduction, a significant decline of 11 crimes per week (11%) at a one-mile radius around the stations, and about 1 crime per week (32%) over the extremely narrow base of crime at the station itself. We also find no evidence of crime displacement. This study first utilized a traditional difference-in-differences methodology, but we also used a new tool, the causal random forest. Both methodologies are compared and contrasted with an eye toward user understanding. The results of our analysis are consistent and coherent across all the different methodologies, with the causal random forest finding more pronounced effects by taking into account two major factors: the propensity of the regions for treatment, and the interaction between elements of interest.
|
|---|