Skip to content

Lukee563/Congestion-Pricing

Repository files navigation

NYC Congestion Pricing and Its Effects on Motor Vehicle Collisions


Authors

  • Luke Catalano - M.S. Quantitative Economics & Finance, UC Santa Cruz
  • Eleanor Stoever - B.S. Applied Mathematics, B.A. Environmental Studies & Economics, UC Santa Cruz

Overview

This paper explores the causal effect of New York City’s Lower Manhattan congestion pricing policy on traffic collisions in the district of Manhattan. The study utilizes a Difference-in-Differences (DiD) empirical approach to evaluate how the policy, implemented on January 5th, 2025, affected total weekly collision counts.

Key Findings

  • Significant Reduction: We find a statistically significant decrease of approximately 20 total collisions per week in the treatment area.
  • Impact Scale: This represents a 16.66% reduction from the average weekly collision count over the 10 weeks following implementation.
  • Economic/Safety Implications: The study focuses on "reportable" collisions—those involving at least $1,000 in vehicle damage or personal injury.

Methodology

The analysis compares two geographically similar areas that exhibited parallel collision trends prior to the policy:

  • Treatment Group: Lower Manhattan (the "Congestion Relief Zone" below 60th Street / 40.7650° N).
  • Control Group: Upper Manhattan (above 40.7650° N).

Empirical Strategy

The study employs a linear regression model with robust heteroscedastic standard errors to account for non-constant variance over time:

$$crash.count = \beta_0 + \delta_0(post_i) + \beta_1(treat_i) + \delta_1(post_i \times treat_i)$$

  • Validation: The "Parallel Trends" assumption was verified through visual trend plotting and a Treatment-Control Balance Table, confirming no statistically significant differences in pre-treatment characteristics such as time of day or injury severity.

Data

The dataset is derived from the City of New York’s Motor Vehicle Collisions-Crashes database, maintained by the NYPD.

  • Period: January 2023 – March 2025.
  • Scope: 115 weeks of data, providing 230 total observations (treatment vs. control pairings).
  • Variables: Includes collision coordinates (latitude/longitude), injury/death counts, and contributing factors.

Project formatted in LaTeX - November 2025

About

Casual Inference of NYC Congestion Pricing on Collision Rates

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages