Urban mobility and crime: causal inference using street closures as an instrumental variable

Area of Science:

• Criminology
• Urban Studies
• Data Science

Background:

• Cell phone mobility data is increasingly used in social science to study everyday movement patterns.
• Existing research suggests ambient population, or visitors, in an area predicts crime, but often without establishing causality.
• Past studies frame neighborhood visitor flows predictively, not necessarily causally.

Purpose of the Study:

• To explicitly estimate the causal effect of visitor flows on crime rates using counterfactual terms.
• To address the gap in understanding the causal relationship between neighborhood visitors and crime.
• To investigate the additive effect of visitors on various crime measurements.

Main Methods:

• Utilized two causal inference approaches: conventional two-way fixed effects and a novel instrumental variable approach.
• Employed high-resolution mobility and crime data from New York City for the year 2019.
• Estimated the causal impact of visitors on multiple crime metrics.

Main Results:

• Two-way fixed effects models indicated a significant effect of visitors on a wide range of crime forms.
• Instrumental variable estimates revealed no statistically significant causal impact of visitors on crime rates.
• Large standard errors in instrumental variable results suggest substantial uncertainty regarding the causal effect of visitors on crime.

Conclusions:

• While conventional methods suggest visitors influence crime, advanced causal inference techniques like instrumental variables indicate a lack of statistically significant causal effect.
• The study highlights the importance of employing rigorous causal inference methods to avoid spurious correlations when analyzing mobility data and crime.
• Further research with larger datasets or refined methodologies may be needed to resolve the uncertainty surrounding the causal relationship between visitor flows and crime rates.