Thread by @mostrovs

One would normally want to wait longer to evaluate the effects of major policy changes, but the results of the first three days of Congestion Pricing in New York City are so striking that it is already possible to make some (at least preliminary) conclusions.

(A 🧵)

Below are two screenshots from (shoutout to the amazing @JoshuaMoshes and @BenjaminMoshes)! On the left is the time it took today to cross the congestion zone from North to South at various times of day (red line) vs. historical average (blue line). congestion-pricing-tracker.com

Basically, no effect: red line tracks the blue one very closely. Same conclusion for all three days (Sun-Tue) on Routes 1-5 on congestion-pricing-tracker. By contrast, there is a clear and significant effect in the screenshot on the right: traffic in Holland Tunnel has improved.

One can again go to congestion-pricing-tracker and see similar substantial positive effects for many of the bridges and tunnels entering Manhattan (Routes 6, 7, 9, 10, 11, 12, 13) on all three days. So why is there such a striking difference?

Key difference is that Routes 1-5 are within the congestion zone, while the bridge/tunnel routes just enter it. Why does it matter? It matters because of taxis and for-hire vehicles (FHVs - Uber and Lyft). These vehicles, by their nature, enter the area and take many trips there.

By contrast, personal vehicles are much more likely to just take two trips per entering (one trip to enter, one trip to go back home). So the routes within the congestion pricing zone have a much higher fraction of taxis/FHVs than do bridges and tunnels leading to it.

Frank Yang and I have hired several assistants to take videos of passing cars in Manhattan and classify them (in New York, you can tell FHVs apart from regular cars by the license plates that say "TLC" on them). Below are two 10-minute videos to give a sense of differences.

Both videos are from Dec 7. First one is from Holland Tunnel exit. Over the 10 minutes, the video shows 179 regular cars, 93 TLCs, and 7 taxis (plus a bunch of cars without front plates - but that's a separate issue). So - a substantial majority of vehicles are regular cars.

The second video is from an intersection within the congestion pricing zone (8th Avenue and 30th Street). Over the similar 10-minute period, the fractions of different types of vehicles are completely different: only 48 regular cars, vs. 60 TLCs and 50 taxis.

Why does this difference in composition matter, and why does it explain the difference in the effectiveness of the congestion pricing plan in the different areas (and the difference between the effects in the two screenshots)?

The answer is in my paper with Frank: . The current plan charges a high amount per trip to regular cars, while charging little to the passengers of taxis and FHVs. So in the areas where the latter type of traffic is predominant, the plan is ineffective.web.stanford.edu/~ost/papers/ny…

This suggests a straightforward fix: charge more to taxis and FHVs (during peak times). In the paper, we work out the specific numbers, and also discuss the related issue of delivery services. We propose the same $9 toll per FHV/taxi trip, or even better, $4.50 per mile.

Having said all this, the current plan is an amazing first step in the right direction - congratulations to all the policymakers for the incredible work they put into making this happen! It's great that it is already working so well on the bridges and tunnels!