Abstract
We study the impacts of New York City's Central Business District (CBD) Tolling Program, the first cordon-based congestion pricing scheme in the United States. Using a generalized synthetic controls approach that compares outcomes in NYC to contemporaneous outcomes in other cities, we find that the policy increased speeds on CBD roads by 11%, with little-to-no effect on air quality, transactions at shops and restaurants, or overall foot traffic in the CBD. Speeds also increased outside the CBD, especially on roads commonly traversed by drivers traveling to/from the CBD. These spillovers lead to faster trips throughout the metro area, including for many unpriced trips. We develop a simple model to bound the effects on driver welfare. If drivers have a Value of Travel Time (VOTT) of $40/hour, then we estimate that driver welfare increased by at least $14.3 million/week, before any revenue recycling or environmental benefits. Passenger vehicles headed to the CBD are only better off if their VOTT exceeds $153/hour, but the modest speed improvements for the many unpriced trips reduce the overall 'break-even' VOTT of these drivers to at most $21/hour. Finally, we show how characteristics of local travel patterns and road networks can inform the potential impacts of introducing cordon-based congestion pricing in other cities.
Introduction
Congestion pricing is a classic Pigouvian tax designed to internalize the externalities of traffic. However, real-world implementations are often "second-best" solutions that price only specific zones. This study examines the January 2025 implementation of New York City’s Central Business District (CBD) Tolling Program to understand how such a policy reshapes urban mobility and economic activity. A central question is the nature of "spillover" effects: does pricing a central zone merely divert traffic to untolled roads, or does it improve flow across the entire network?
Methodology
To isolate the impact of the tolling program, we employ a Generalized Synthetic Control (GSC) design. We compare NYC’s traffic and economic data against a "synthetic" NYC constructed from five control cities: Philadelphia, Boston, Chicago, Atlanta, and Baltimore. This methodology allows us to account for seasonal and national trends that might otherwise confound the results.
Our analysis integrates several high-granularity datasets:
- Mobility Data: Google Maps Traffic Trends for road speeds and Advan Research for GPS-based foot traffic.
- Economic Indicators: MBHS3 credit/debit card transaction data covering approximately 180 million individuals to track retail and restaurant spending.
- Environmental Data: PM2.5 concentrations from over 1,000 PurpleAir sensors.
- Network Modeling: Replica mobility simulations and Bureau of Public Roads (BPR) congestion functions to map the relationship between traffic density and speed.
Key Findings
Traffic Speeds and Network Spillovers
The policy successfully reduced congestion within the priced zone, with CBD road speeds increasing by 11% (from 7.1 mph to 8.2 mph). However, the benefits extended far beyond the Manhattan cordon. We find significant "spillover" speed increases on roads outside the CBD, particularly on highways leading into the city.
Speeds on these highways increased by 13.5%, a larger gain than that seen on local CBD streets. This discrepancy is likely due to "auxiliary congestion" within the CBD—factors like double-parking, pedestrian crossings, and frequent intersections—which keep speeds low even as vehicle volume decreases. Because NYC roads were operating at highly elastic points on their congestion functions, even modest reductions in volume yielded outsized gains in travel speed.
Economic and Environmental Impacts
Contrary to concerns that the toll would deter commerce, we find no detectable drop in retail transactions, restaurant spending, or overall foot traffic within the CBD. Travelers appeared to adapt by switching to public transit or carpooling rather than canceling trips entirely.
Environmentally, the policy led to an estimated savings of 653 tonnes of CO2 per week due to improved fuel efficiency. Interestingly, there was no measurable change in PM2.5 concentrations. This suggests that in modern American cities with stringent vehicle emission standards, congestion pricing may be more effective as a time-management tool than as a primary strategy for reducing localized air pollution.
Welfare Analysis
We estimate the net benefit to drivers using a welfare model based on the Value of Travel Time (VOTT). Assuming a VOTT of $40/hour, the policy generated at least **$14.3 million per week** in net driver welfare gains.
A critical insight is the distribution of these gains. While drivers entering the CBD and paying the toll only "break even" if their VOTT exceeds $153/hour, the aggregate welfare is bolstered by millions of **unpriced trips**. Because these unpriced trips share road segments with CBD-bound traffic, they benefit from the reduced volume without paying the toll. When accounting for these diffuse time savings, the overall break-even VOTT for the driving population drops to just $21/hour.
Conclusion
The short-run evidence from NYC suggests that cordon-based congestion pricing can significantly improve network-wide traffic speeds without harming local economic activity. The primary beneficiaries are often those making unpriced trips that traverse shared infrastructure. Future research should address long-term structural adjustments, such as changes in residential location or business relocation, and the further welfare gains realized through the recycling of toll revenues into public transit infrastructure.
Summary updated