What congestion pricing saved everyone on the road
Zone: Manhattan below 60 St, pricing since 2025-01-05. Run 2026-10-05.
The bus note valued only bus riders’ time (about 525,000 hours, $10 million a year). That leaves out the people in cars, taxis, for-hire vehicles and trucks, who are most of the traffic. This note estimates their time.
Short answer
Section titled “Short answer”Road users in the zone saved about 5 million person-hours in 2025, worth about $110 million at federal values of time (range $39 to $251 million). With bus riders, the total is about 5.6 million hours and $120 million (range $45 to $267 million). That is the time saved in the zone only. It leaves out any gain on roads outside the zone, which a study using Google Maps data finds is larger (see “Other estimates”).
| Per year, 2025 | Low | Central | High |
|---|---|---|---|
| Speed gain in the zone | +1.5% | +3.1% | +4.0% |
| Vehicle-hours saved | 1.5 million | 3.4 million | 5.0 million |
| Person-hours saved, traffic | 2.0 million | 5.1 million | 7.5 million |
| Value, traffic | $39 million | $109 million | $251 million |
| Bus riders (from the bus note) | $6 million | $10 million | $16 million |
| Value, total | $45 million | $120 million | $267 million |
| Person-hours, total | 2.3 million | 5.6 million | 8.3 million |
The central figure has one number you can check by hand: 112 million vehicle-hours spent in the zone in 2025, times a 3.05% speed gain, is 3.4 million vehicle-hours.
The data
Section titled “The data”All from the MTA on data.ny.gov.
| Dataset | What it holds | What it lacks |
|---|---|---|
jbxx-kxff vehicle miles traveled, 2024 on | Average daily vehicle-miles in the zone (CRZ) and on excluded roads (FDR, West Side Hwy), by month, Jan 2024 to Jun 2026. Two estimates: from aggregated GPS (agps) and from connected-vehicle data (cvd). | No hours of the day, no vehicle class. The two models differ by 20–30% in level (2025 zone average: 3.0 million vs 2.4 million miles a day) and in the change (below). |
6p29-6xqn taxi and for-hire speeds, Oct 2019 on | Average mph, by month, in the CBD, CBD Adjacent and Greater NYC (from TLC trip records). | Monthly only: no weekday/daytime split. Ends August 2025, so nothing for 2026. |
t6yz-b64h zone entries, 2025 on | Entries by crossing, vehicle class and 10-minute block, 5 Jan 2025 to 26 Sep 2026 (6.5 million rows). | Starts with pricing: no pre-pricing comparison inside the dataset. Entries only; vehicles already inside are not counted. |
Why taxi and for-hire speeds stand in for all traffic. The MTA publishes no all-vehicle speed series for the zone. Taxis and for-hire vehicles are about 38% of entries and drive in the same lanes as everyone else. Their speed includes pickups and drop-offs, so their trips run slower than a through car; what we use is the percent change, which should carry over. Check: the MTA’s own all-vehicle figure for the first year is weekday speeds up 4% (Governor’s release, below); ours is 3% to 4%.
Caveats on the data: 2025 vs 2024 months are matched by calendar month; the zone’s VMT is all hours and all days, so we apply an all-hours speed change to all-hours vehicle-hours (no daytime split is possible); January 2025 includes four days before pricing.
Method and results
Section titled “Method and results”1. Speed change. For each of the eight months with data (Jan–Aug), the log change in CBD speed from 2024 to 2025; the mean, with a 95% interval across the eight months (t distribution).
| Same months, Jan–Aug | Change | 95% interval | Months up (of 8) |
|---|---|---|---|
| CBD, 2025 vs 2024 | +3.1% (8.73 to 9.01 mph) | +1.5 to +4.7 | 7 |
| CBD, 2024 vs 2023 (placebo) | −3.5% | −4.3 to −2.7 | 0 |
| CBD, 2023 vs 2022 (placebo) | −4.9% | −7.1 to −2.6 | 0 |
| CBD Adjacent, 2025 vs 2024 | −1.1% | −2.9 to +0.7 | 2 |
| Greater NYC, 2025 vs 2024 | −0.9% | −1.9 to +0.1 | 1 |
| CBD minus Greater NYC, 2025 vs 2024 | +4.0% | +2.8 to +5.1 | 8 |
- Speeds in the zone fell in each of the two years before pricing (−4.9%, −3.5%) and rose in 2025. A reversal of that size is not noise. The two before-years are a downtrend, not a flat baseline, so measuring against 2024 is the cautious choice: against the trend, the 2025 gain would be larger.
- Citywide speeds fell about 1% in the same months, so taking that out gives +4.0%, the “high” case and the MTA’s own figure.
- Month by month: up in every month but July (−0.3%); biggest in March (+5.9%), January (+3.8%) and August (+4.1%). See
congestion_pricing_traffic_speeds_monthly.csv. - Adjacent areas did not slow in a way that stands out (−1.1%, interval spans zero), so we see no sign of traffic pushed just outside the zone.
- Not possible here: a weekday-daytime split, and 2026 (the speeds end Aug 2025). The bus note’s 2026 fade suggests the 2025 figure overstates 2026.
2. Vehicle-hours. Hours the 2025 traffic took in the zone: monthly vehicle-miles × days ÷ 2025 speed (112 million hours; about 308,000 a day). Hours saved = those hours × (speed 2025 ÷ speed 2024 − 1), which equals VMT₂₀₂₅ × (1/speed₂₀₂₄ − 1/speed₂₀₂₅). Jan–Aug use each month’s measured speeds; Sep–Dec (no speed data) assume the same percent gain. Jan–Aug alone is 2.2 million vehicle-hours.
The counterfactual is explicit: the same 2025 traffic at 2024 speeds. We hold the traffic fixed and ask how much faster it moved.
- Low:
cvdVMT and the bottom of the speed interval (+1.5%). Central: the average of the two VMT models and the measured speeds (+3.05%). High:agpsVMT and the +4.0% from the Greater NYC comparison.
Fewer vehicles is a separate item, not a saving. VMT in the zone was 3.8% lower in 2025 than 2024 by agps and 11.0% lower by cvd (2026, Jan–May: −8.5% and −14.5%). The cvd fall matches the MTA’s 11% fewer entries; the agps fall is smaller. Those vehicle-miles would have taken another 3 to 8 million vehicle-hours at 2024 speeds (Jan–Aug 2025). They are trips that drivers chose not to make or made another way. Those people are not better off by the hours not driven (they gave up something, or paid with a toll or a transit trip), so we do not count the hours. The only part counted is that traffic that remained moved faster. (Cook et al. take the same line: those who stop driving lose no more than those who stay.)
3. People and dollars.
| Input | Value | Source |
|---|---|---|
| Passenger vehicle occupancy | 1.34 low, 1.52 central and high (all travel). Others in the table: weekday off-peak 1.41, weekend 1.81 | USDOT BCA Guidance, May 2025, Table A-3 (2022 National Household Travel Survey), p. 40 |
| Value of time, passenger | $19.40 personal (low), $21.10 all purposes (central), $33.50 business (high; every trip on business, a ceiling) | Same guidance, Table A-2, 2023 dollars per person-hour, p. 39 |
| Trucks | One driver at $35.70 an hour (operator only; no freight value) | Table A-2, “Truck Drivers” |
| Class mix | 94.5% passenger (cars, taxis, FHVs, motorcycles), 3.9% trucks, 1.6% buses of 2025 zone entries | Entries dataset |
Buses are dropped from the traffic part, since their riders are counted separately. The class mix is by entries, not by vehicle-hours (taxis and FHVs circulate inside the zone, so their true share is higher; trucks stay a small share). It barely matters: a passenger vehicle-hour is worth 1.52 × $21.10 = $32.07 and a truck-hour $35.70.
The USDOT guidance’s local personal value ($19.40) is for local trips of all sorts. Manhattan wages are above the US average, so $19–$21 is on the low side for this place. Cook et al. use $40 an hour (80% of the NYC-area wage). At $40 for all, our central traffic value would be about $200 million.
4. Add bus riders. The bus note’s figures: 525,000 rider-hours (range 293,000 to 837,000), $10.3 million (valued at the earlier $19.60 USDOT figure, kept as is). No double count: buses are 1.6% of entries and are excluded above.
Other estimates
Section titled “Other estimates”| Source | Figure | How it compares |
|---|---|---|
| MTA / Governor, first anniversary (Jan 5, 2026): release | Weekday speeds in the zone +4%, weekends +6.2%, trucks +5.6%, buses +2.3%. 27 million fewer vehicles in the year, 73,000 a day, 11%, from a base of about 664,000 a day. Net revenue $518 million through November 2025, over $550 million projected for the year. | Their weekday speed gain matches our high case (4.0%). No time-savings figure given. |
| Our entries data | An average of 494,000 entries a day in 2025 (Jan–Mar 481,000, Jun–Aug 492,000), 2026 Jan–Sep about 470,000; the dataset starts with pricing. | No pre-pricing baseline in the dataset; the MTA’s 664,000 is its own estimate for 2024, and its release does not say how it was built. |
| Cook et al., NBER 33584 (Aug 2026; Google Maps speeds, synthetic control) | Speeds on road segments in the zone +11% (+15% in the first four months, +10% after). Gains also outside the zone. Driver welfare gain of at least $7.5 million a week (about $390 million a year) at $40 an hour, net of tolls paid. About $17.3 million a week comes from 180 million unpriced weekly trips, which each save 9 seconds. | Different scope and measure, see below. |
Why Cook et al. is much larger than ours. Three differences, in rough order of size:
- They count time saved on trips outside the zone: $17.3 million a week, about $900 million a year, which at $40 is about 22 million person-hours a year across the metro area. Ours is zone-only.
- They value time at $40 an hour; we use $19–$34.
- Their speed measure (segments, Google Maps users) rises 11%; the MTA’s taxi/FHV speeds rise 3 to 4%. Google Maps users follow routes through a segment and weight slow segments differently from a taxi trip’s average speed.
Their figure is a lower bound on net welfare, after subtracting the tolls drivers pay, so it is not “time saved”. The toll itself is a transfer to the MTA; it is counted separately from our figures, which are time only. Put side by side, our zone-only 5 million hours is a fifth of their metro-wide 22 million, and both sets of numbers agree on the shape of the result: most of the time saved comes from a small speed gain spread over very many trips.
Caveats
Section titled “Caveats”- Speed measure. Taxi and FHV speeds stand in for all traffic. If private cars gained more or less, the dollar total moves in proportion.
- Annual figure. Speed data stop in August 2025. Sep–Dec 2025 use the Jan–Aug percent gain. The bus note and Cook et al. both show the gain fading over time; 2026 is likely lower.
- VMT. The two MTA models disagree on level and on change; we show both ends. No hour of day or class split, so no peak or off-peak breakdown.
- Occupancy and class. National occupancy applied to a Manhattan mix that includes many taxi and FHV vehicles. Taxis have empty cruising time, which lowers true occupancy; delivery vans and commuters raise it.
- Value of time. Federal figures, low for Manhattan. At $40 the central estimate is about 1.9 times higher.
- Not in the total: gains outside the zone; reliability (fewer extreme delays, per Cook et al.); emergency vehicles (one study finds about a minute off ambulance trips); vehicle operating costs; safety and air quality; trips not made; the toll paid. Fewer trips and the toll are costs to some people, so this is time saved, not the net benefit to drivers.
- Cause. Pricing is the main change in the zone in 2025, and the before-years show speeds falling, but the speeds are not randomized: roadwork, weather and ridership shifts move them too.