Why Traffic Congestion Has Rebounded in the CBD

Traffic congestion in the Manhattan core is rebounding. Travel-speed data culled from taxicab GPS and released last month by City transportation and taxi officials suggest that average motor vehicle travel speeds in the Central Business District fell by 8.5 percent from 2012 to 2014. The slowdown follows years of flat or even rising speeds — a phenomenon that predated the 2008 financial collapse and undermined congestion pricing proposals by making car, bus, truck and taxi travel in the heart of the city a little more efficient and predictable.

The speed data were issued during the recent Uber controversy, but they bear on a host of other policy issues including Vision Zero, cycle infrastructure, and bus priority signals and lanes. To inject some rigor into these discussions, I’ve created a special section of my Balanced Transportation Analyzer spreadsheet to approximate various factors’ roles in re-congesting the CBD. My preliminary findings are summarized in the graphic. (Readers interested in a deep dive should download the BTA and navigate to the new tab, CBD Congestion.)

Table of Congestion Causation _ 28 July 2015
Graph: Screen shot from the Balanced Transportation Analyzer

The task is to account for an “adjusted” 10.5 percent drop in CBD travel speeds. That figure tacks 2 percentage points on top of the observed 8.5 percent reduction to reflect the fact that the 20,000 a day decrease in motor vehicle entries from 2012 to 2014 also reported by DOT should have increased CBD travel speeds by 2.0 percent. Here are details:

  • Uber: My calculation that the presence of Uber vehicles in the CBD is reducing travel speeds by 4.3 percent, is predicated on 50 percent additionality, with half of Uber CBD pick-ups constituting additional traffic while the other half displace yellow cabs and traditional black car service. (Additionality is almost certainly higher for Uber trips outside the CBD, where yellow cabs are far less prevalent.) I previously published a higher estimate of Uber’s impact, but that was predicated on 75 percent additionality. The 50 percent rate used here is based on data-mining by the New York Times, published yesterday in its “Upshot” column.
  • CBD Cycle Infrastructure: Bicycle lanes draw some mode share from taxis and app-based for-hire vehicles, and they arguably make traffic flow more orderly, but on balance the space they take from motor vehicle use probably outweighs those effects. I estimate that CBD bike lanes added from 2012 to 2014 effectively removed eight-tenths of one percent of available CBD street space, which, by my modeling, would have led to a 1.9 percent average slowdown in motor traffic.
  • Cheaper Gasoline: Pump prices have fallen around a dollar a gallon since last fall, cutting an estimated 6 percent from the cost of a typical automobile round-trip to the CBD, which my modeling suggests would lead to 3 percent more trips, or 20,000 additional CBD auto trips a day. (Note that this increase, like the others reported here, is a net figure that assumes that incremental gridlock causes some other auto trips to disappear.) While this rise would postdate the latest official figures on CBD vehicle entries, I’m betting it will be reflected in a modest uptick in 2015 data. The associated reduction in CBD speeds is 2.0 percent, according to the BTA.
  • Pedestrian Traffic and Vision Zero: Three possible phenomena intersect here: more people on foot in the CBD, greater tendency of turning drivers to wait for pedestrians to clear crosswalks, and the 2014 local law reducing the speed limit on NYC streets to 25 mph (from 30 mph). I give credence to the first two categories and have assigned them modest vehicle speed reductions, 0.5 percent each. But I grant only a de minimis 0.1 percent effect to the third, based on the infrequency of opportunities to reach 30 mph in the CBD as well as lackadaisical NYPD enforcement of the new lower limit.

Summing up, my half-dozen factors account for 9.3 percent of the adjusted observed 10.5 percent drop in CBD travel speeds. The remainder could be due to Uber (I may have lowballed the traffic impact of their deadheading vehicles), or the advent of other app-based FHV’s such as Lyft, or errors in any of my component estimates. This is not an exact science, and my decimal-point figures are only meant to convey central estimates and not to imply precision.

What’s the bottom line? Clarity, I hope. Public policy works better when it’s built on informed analysis. Still, parsing the resurgence in traffic congestion isn’t nearly as vital as recognizing and responding to it as the calamity it is. There’s still one proposal out there for alleviating CBD gridlock that can also finance the new wave of transit upgrades and reform the city’s inequitable tolling system. And, unlike the 2007-08 congestion pricing proposal, there’s just one man standing in the way.

  • roguebagel

    So a bike lane, which reduces the width of a vehicle lane by 3 feet, is responsible for half of the congestion created by Uber? That seems absurd. What about lane reductions without the presence of bike lanes – that surely must equal more space than the increased bike lanes put together, if streetsblog and others are to be believed.

    And isn’t this conflating “congestion” with “average vehicle speeds”….again?

  • marks

    Presumably, the effect of bike lanes could be estimated by looking at speeds in the one year before & after installation, for each installation. If these effects were averaged across a number of installations, it would help control for other effects such as Uber’s ramp-up.

  • Mathew Smithburger

    We are not attributing enough weight to cheaper gas prices combined with larger vehicles and increased construction.

  • Jesse

    Bike commuters should not be uncomfortable with the idea that bike lanes might contribute to congestion. From a driver’s perspective, everything in the city that’s not asphalt dedicated solely to private cars (except of course your car) contributes to congestion. If the streets were cleared of bike lanes, bus lanes and uber, drivers would complain that sidewalks cause congestion. Parks cause congestion. Buildings cause congestion. Cities cause congestion. There is no arguing with the statement that bikes cause congestion because the premise is wrong. From the motorist’s point of view, everything causes congestion but cars.

  • roguebagel

    It’s based on an assumption that all bike lanes contribute to congestion because you can sum up the total miles of lanes and take away that many vehicle lanes once you factor for width (the author assumes 50%, which is interesting to say the least). The impact is overstated.

  • Komanoff

    I appreciate your diving into the BTA to try to figure out how I treated bike lanes. I’m sorry you stopped at 2006-11, though. That was just the backdrop for 2012-14, when fewer lanes were added.

    I didn’t crunch actual width of CBD bike lanes. I should. It may also be that CBD neighborhoods with bike lanes were less congestion-prone than average. Let me know if you have info on either front.

    I share some of your concern that I’ve charged bike lane expansion with 45% of the congestion causation of Uber. Yet that’s where my numbers take me. Do you think my Uber # is too low or my bike lane # too high, or both?

    I welcome critiques, especially with specifics/data. Also, M. Smithburger, let me know if you have data on the pace of construction in the CBD. Thanks.

  • Bobberooni

    The bike lanes were put in by a data-aware Bloomberg administration that sought to put them in places that were not bottlenecks, thus having a minimal impact on congestion. Before-and-after measurements bore this out, and in fact showed traffic speeds increasing in some cases. Eliminating Broadway as a major thoroughfare had a particularly good affect on INCREASING traffic speeds.

    The requirement that bike lanes are ONLY installed in places where they won’t slow down traffic is probably one big reason most of them kind of peter out when you reach the CBD.

    So no, I don’t believe this analysis.

  • Joe R.

    It’s worth noting when looking at this analysis that even the broader trends in vehicle speeds up or down aren’t that dramatic. Whether average vehicle speeds are 7 mph or 9 mph, they’re still dog slow, and NYC is still paying dearly for this in terms of lost economic activity. Had the fall been from 20 mph down to 9 mph then we would be talking a huge impact. The bottom line is NYC won’t be competitive until traffic is free-flowing most of the time at a reasonable speed. Add in more smartly timed signals and lots more roundabouts, then you could be talking average speeds over 20 mph. But we’ll never get there unless we charge enough to use roads so only essential traffic considers it worthwhile to pay the fee. Charge a fee to drive anywhere traffic currently isn’t free-flowing for most of the day (that will also include parts of the outer boroughs). Keep increasing that fee until traffic is free-flowing.

    This is a great analysis but it merely underscores a fact NYers already know—it’s slow getting around the CBD. Despite slight fluctuations up or down, it has been slow for decades.

  • Joe R.

    One thought just occurred to me which these data don’t address. What is the impact on CBD vehicle speeds to vehicle speeds in the outer boroughs? I tend to think there is a huge compounding effect where a small decrease in CBD vehicle speeds might cause a huge decrease in vehicle speeds elsewhere. Here’s why. The Manhattan CBD is already near saturation with vehicles. 5% or 10% more traffic might cause a single digit percentage decrease in traffic speeds in order to accommodate it because the streets are already well above the traffic levels needed for free flow conditions. However, much of Manhattan traffic originates outside Manhattan. Now it’s well known that roads have a breaking point where a few percent more traffic can cause free flow conditions to break down. How many outer borough roads reach this condition due to a few percent more vehicles driving into the Manhattan CBD? Anecdotally it seems to me traffic congestion is much worse local to me compared to a few years ago. The Q64 to the subway often takes 20 to 25 minutes, whereas it took about 15 minutes a few years back. A lot of the arteries crawl along for much of the day. Doubtless other areas have experienced this also.

    I don’t have the data to give a good answer here but it’s something I thought worth mentioning. It may turn out congestion pricing not only improves things quite a bit in Manhattan, but that it improves them dramatically in the outer boroughs.

  • Alexander Vucelic

    Re bike lanes

    Some percentage of bike trips in the CBD would otherwise be taken in motor vehicles. For example; my wife & I used to frequently take cabs, now rarely.

    The number of daily bike trips in the CBD runs roughly between 100,000 and 200,000 per day. This is admittedly a crude estimate based on Citibike volume plus a rough addition for private cyclists and some accurate traffic counts.

    Also note, bikes are a significant percentage of CBD road traffic these days. For example, last week during early rush hour at 45th and Fifth Avenue, 19% of vehicles using the roadbed were bicycles; despite zero bike infrastructure, and harrowing conditions. Traffic Counts throughout the CBD indicate cyclists are anywhere between 7% and 40% of roadway traffic.

    note, that the DOT bike counts woefully undercount rider traffic. For example, the DOT does not count bikes inside the CBD. The DOT does not count ‘commercial’ riders.
    Some protected bikes lanes in the CBD currently see 500-600 riders at peak hour. These 500-600 peak hour riders are easily accommodated. These are 6′ lanes. In stark contrast, 12′ motor vehicle lanes struggle to reach 500 vehicles at peak hour.

    Joe R. had calculated that protected bikes lanes have 10x the carrying capacity of motor lanes.

    For more insight into recent traffic counts – contact Janet Tiff over at TA.

    Your model needs to estimate some number of cars removed from CBD due to increased bike riding.

  • BBnet3000

    I’m not aware of any bike lanes that took space cars were actually using.

    That’s why they are only on peripheral Avenues and even then have skipped Midtown. They improved travel speeds on those Avenues because while taking away unused capacity they also removed most vehicles waiting to turn from the traffic flow.

  • Alexander Vucelic

    one should also recognize that there are bike lanes and then there are bikes lanes.

    sharrows – are effectively not a bike lane. Try riding 2nd Avenue from 60th to 32nd. motors use these often ‘sharing’ the lane with riders

    painted bike lane – a kinda sorta bike lane. motors will use these

    protected bike lane – indeed these are semi-exclusive bike lanes which BTW do not remove one single motor lane in the CBD

  • ahwr

    Has NYCDOT published throughput or travel time numbers before/after for most bike lane installations? If they have, for those time periods are there any CBD-wide numbers you could compare them to?


    There’s a short ‘mobility’ section that might be worth a look, but isn’t as comprehensive as I was hoping for.

    I had thought the bike lanes were placed mostly where the impact on traffic would be minimal. Since the bike lane would be taking up what had been wasted or lightly used space, or there were already a large number of bikes and segregating them would improve flow for motor vehicles. So the amount of travel time delay you attribute to bike lanes seems high.

    Shouldn’t there by a line item for bike traffic though?


    Maybe bike lanes have had such a large impact, but most of the impact has been indirect, by increasing bike traffic on roads without bike lanes. If the city has taxi GPS data would they be willing to break down where speeds have decreased fastest? A correlation with increased bike counts could support a congestion reduction plan that included a targeted expansion of bike facilities.

    (if you dealt with this in your excel file, I wasn’t able to load it in google’s spreadsheet program or open office)


Wanted: Crowd-Sourced Transportation Analysis

My recent post refuting David Owen’s attack on congestion pricing ignited a long, rich thread. Here’s one comment, from "Jonathan," that struck a nerve: [A] cordon-pricing plan … which doesn’t charge center-city residents could result in an increase in those residents’ automobile use. If the streets are free of outer-borough traffic, more of my Manhattan […]

What Can Taxi Data Tell Us About NYC Streets?

The average density of taxi pick-ups at 1 a.m. on Saturdays in 2009. The most rides originated from the Meatpacking District and the Lower East Side. Image: NYT. When New York City installed GPS units in its taxi fleet in 2007, it began an ambitious initiative to gather information about how traffic functions. Over the […]