Skip to content
AS / SIGNALS
Back to work

Policy Research

Subway Fare-Evasion Arrest Intensity — Race and Poverty

Station-level analysis of Brooklyn fare-evasion arrests: a one-percentage-point higher area poverty rate is associated with 4.61 more arrests per 100,000 swipes; high-poverty stations average 1.42 vs. 0.78.

Role: Analyst (Columbia SIPA DSPC 7514)

Situation

Whether fare-evasion enforcement intensity tracks neighborhood poverty and racial composition is a live NYPD/MTA policy question. Station-area demographics are a noisy proxy for who rides, so results need explicit caveats.

Decision question

Station-level analysis of Brooklyn fare-evasion arrests: a one-percentage-point higher area poverty rate is associated with 4.61 more arrests per 100,000 swipes; high-poverty stations average 1.42 vs. 0.78.

Data and constraints

See method and results below for sample, coverage, and documented limits.

Method

Aggregated cleaned arrest microdata to stations, joined 2016 ridership and neighborhood poverty/race data, dropped Coney Island as a tourist-ridership outlier, defined arrest intensity as arrests per 100,000 swipes, and estimated ridership-weighted HC2-robust regressions of intensity on poverty, high-poverty status, and majority-Black station areas.

Tools

  • R
  • tidyverse
  • estimatr (lm_robust)
  • modelsummary
  • ggplot2

Deliverables

  • 18-page knitted analysis (February 2026)
  • Station-level joins, intensity metrics, and weighted regressions

Outcome / decision value

  • A one-percentage-point increase in station-area poverty is associated with 4.61 more arrests per 100,000 ridership (p < 0.001; R² ≈ 0.15).
  • High-poverty station areas average 1.42 arrests per 100k swipes vs. 0.78 in other areas (difference 0.63, p < 0.01).
  • Highest-intensity stations (Junius St, Atlantic Av, Livonia Av, Sutter Av) are predominantly majority-Black, high-poverty areas. Cross-section cannot separate enforcement from underlying fare evasion or deployment.

Links

GitHub URL — add in content-data.tsLive demo URL — add in content-data.tsAnalysis PDF