Trips into the congestion zone

Congestion pricing started on January 5, 2025, for vehicles entering Manhattan below 60 St. Taxis and app cars pay a per-trip fee there, recorded as cbd_congestion_fee. Here we use that fee to find the zone, then count trips a year apart.

This is a short version of our taxi and app study, which adds speeds and intervals.

Step 1: find the zone

The TLC’s 263 taxi zones don’t line up with 60 St, but the fee tells us which zones are inside. A trip that starts and ends in the same zone inside the area should nearly always carry the fee; one outside, almost never.

ATTACH 'https://data.transitlab.nyc/v1/transitlab.duckdb' AS transitlab (READ_ONLY);

CREATE TABLE cbd AS
SELECT pickup_zone_id AS zone, any_value(pickup_zone) AS name
FROM transitlab.taxi_yellow_trips
WHERE pickup_datetime >= '2025-03-01' AND pickup_datetime < '2025-04-01'
  AND pickup_zone_id = dropoff_zone_id
GROUP BY zone
HAVING count(*) >= 100 AND avg((cbd_congestion_fee > 0)::INT) >= 0.6;

SELECT count(*) FROM cbd;  -- about 37 zones

Zones that straddle 60 St land between 5% and 60%. They are left out here, so trips to Central Park or Lenox Hill count as neither in nor out.

Step 2: count trips a year apart

WITH trips AS (
  SELECT 'yellow' AS kind, pickup_datetime AS t, pickup_zone_id AS pu, dropoff_zone_id AS dz
  FROM transitlab.taxi_yellow_trips
  WHERE pickup_datetime >= '2024-02-01'
  UNION ALL
  SELECT 'uber/lyft', pickup_datetime, pickup_zone_id, dropoff_zone_id
  FROM transitlab.fhv_trips
  WHERE pickup_datetime >= '2024-02-01'
)
SELECT kind,
       year(t) AS year,
       count(*) FILTER (WHERE pu IN (SELECT zone FROM cbd) OR dz IN (SELECT zone FROM cbd))
         / count(DISTINCT t::DATE) AS touching_zone_per_day,
       count(*) FILTER (WHERE pu NOT IN (SELECT zone FROM cbd) AND dz NOT IN (SELECT zone FROM cbd))
         / count(DISTINCT t::DATE) AS elsewhere_per_day
FROM trips
WHERE month(t) BETWEEN 2 AND 8           -- the same months each year
  AND dayofweek(t) BETWEEN 1 AND 5       -- weekdays
  AND hour(t) BETWEEN 6 AND 19           -- 6 am to 8 pm
GROUP BY kind, year
ORDER BY kind, year;

January is left out: pricing began partway through it. Comparing the same months each year keeps the seasons out of the answer.

Reading it

Compare the change in touching_zone_per_day with the change in elsewhere_per_day. Trips can rise or fall citywide for many reasons (fares, the weather, the economy). What pricing might explain is the gap between the two.

This query reads a lot: two years of the two largest tables. Expect a few minutes on a laptop with a good connection. To go faster, keep just the columns used here in a local table first (see Keep queries fast).

Going further

• Speeds. Compare trip_miles / trip_minutes for trips inside the zone with trips entirely above 60 St. The study does this and holds each zone pair fixed.
• Buses. Congestion pricing: buses uses the MTA’s bus speeds inside and outside the zone.