In Any Language, the Cost of Congestion Comes Through Loud and Clear

komanoff_graph.jpgAn analysis using the Balanced Transportation Analyzer shows how much time individual drivers steal from fellow drivers by choosing to drive into the New York City CBD.

It’s not often that you get to see your work set off a Eureka moment for someone else — particularly when that someone is from a different
culture. But I had that experience recently, and it seems worth sharing on
Streetsblog in light of the interest shown today in my analysis of the travel
delay costs from FreshDirect deliveries.

I presented a paper last week at an international forum on
traffic congestion in Guangzhou, China.
People in that city are beginning to look at congestion pricing, and I was asked
to discuss why the Bloomberg toll plan failed politically.

As part of my talk, I described the “social delay costs”
from an additional car trip into the center of Manhattan — literally, the total time that all road users combined spend in traffic because
any one of them decided to drive. Afterwards, one of the organizers, a
professor of transportation engineering, asked me to present a technical version of my paper to his students at South China
University of Technology.

The next day, when I came to the part about social-delay
costs, the professor peppered me with questions about my methodology. As I went
through the steps — basically, every trip takes up an incremental amount of limited street space, which lowers speeds, which adds to everyone’s travel times — the professor
grew more intrigued. It wasn’t that the idea itself was new, but that if
traffic speeds and other baseline data were known, then the delay-impact of one
trip could be quantified. And,
moreover, that the impact varied enormously depending on the time of day: when
there is ample spare road capacity, say, in the middle of the night, an extra
trip has little discernible impact, whereas one trip during congested peak
times adds several hours to the aggregate time that all other vehicles must
spend on the road.

I daresay that for the professor, my elucidation of one
trip’s delay costs helped move congestion pricing from the realm of
abstraction to something tangible and, perhaps, essential. If a peak trip to
the center of New York or some other city can impose one or two hundred minutes
worth of delays on others — and if no driver is ever called on to take that impact
into consideration — then of course the city will be awash in gridlock. No city, not
even Guangzhou, despite an emerging
21st century transit infrastructure of Bus Rapid Transit and new
subway lines, will be able to forestall the tide of free driving.

The same construct animates the FreshDirect analysis in my
Time Thieves paper, except that there the bulk of the delays result from the
trucks’ double-parking. The point is the one I made in my Dot Earth post from Guangzhou: Motorists who pay only for their own lost time, but not for the time their trips
take from other motorists, have little incentive to make efficient decisions about when to drive and how often. In
the case of FreshDirect, this "time theft" averages $15 per delivery. If that
cost were added to the delivery price, FreshDirect’s business, I estimate,
would drop off by around 20 percent.

Then again, no one in New York City — myself included — is proposing congestion tolls even close to the social
delay costs of the trips that would be tolled. The Kheel-Komanoff Plan’s $2-$9
variable tolls ($2-$3-$4 on weekends and holidays, $3-$6-$9 on weekdays) are a
little under 10 percent of the same trips’ respective $30-$130 congestion costs. Yet, as I told the forum in Guangzhou, even this toll — modest relative to the trip’s full social cost — would eliminate enough car trips
that speeds within the Manhattan CBD would rise more than 15 percent.

  • Is there is mislabel in the diagram here? The “outbound” trip seems to cause a delay of 1.7 hours, no matter what the time of day.

  • Ken:

    It’s not a mislabel, just a lack of clarity on my part.

    Outbound trips vary just as much in their congestion causation, from one time period to another, as do inbound trips. The problem is that we don’t know the period in which any trip that has entered the CBD, and paid its toll, will depart.

    If the toll plan is designed to charge only in one direction — which as you know is the case in Singapore, London, Stockholm, etc. and is assumed for NYC — then we’re forced to impute the same “average” outbound congestion causation to all inbound trips, which is what the graphic shows.

    I hope this clears up the confusion. Let us know if you would show this differently (or, even better, would design the toll differently).

  • How does this interact with the proposal for bridge tolls? If the electronic road pricing zone is the CBD plus the Upper East Side and Upper West Side, as proposed by Bloomberg, then this more or less corresponds to your analysis. But if it’s all of Manhattan, as proposed by the otherwise more sensible and more politically palatable plan, then you’re cutting off Harlem from fresh food. I’m not sure the tradeoff of lower pollution versus lower access to fresh food is going to be good for public health.

  • “The problem is that we don’t know the period in which any trip that has entered the CBD, and paid its toll, will depart.”

    But you could model some typical trips. Eg, if someone commutes to Manhattan by car, they come in during the AM Peak and leave during the PM Peak. They would cause more than the delay shown on the graph for someone who comes in during the morning peak.

    I see your point about congestion charging. We just charge people coming in and we don’t know when someone coming in during the AM peak will leave, so these two way models would not help in setting the charge.

    But it would still be interesting to see the two-way congestion cost of someone who commutes into Manhattan for a 9-5 job, of someone who reverse commutes from Manhattan to a 9- job elsewhere, someone who drives in for an evening, and of other typical trips.

  • Downtowner

    The real cost of congestion taxing is documented annually in London by the London Chamber of Commerce and Industry’s annual report since 2005. The LCCI has for all intents and purposes stated that congestion taxing has been the death knell for so many, many small businesses in the pricing zone. Businesses that can’t afford to move out of the pricing zone report regular losses and letting go of employees due to congestion taxing. Will someone please look up the dismal annual reports of the LCCI and display them? This is documented proof of how regressive this tax really is. This is an awful idea for NYC – we’re longtime losing small businesses left and right already. Small businesses are the heart and soul of any large city.

  • Jeffrey W. Baker

    The LCCI appears to be an industry group rather than a credible academic or governmental body.

  • Jeffrey W. Baker

    The LCCI said this only three months ago:

    “Sales, turnover and profitability have increased at bars, pubs and restaurants across the Square Mile, according to a snap survey by the London Chamber of Commerce and Industry (LCCI).”

    The “square mile” is the City of London, which lies entirely within the congestion charging zone.

  • I believe the calculation is only of delays caused within the congestion cordon, and there obviously more delays caused outside of the cordon. For example, if someone commutes from Long Island to Manhattan for a 9 to 5 job, he might cause 5 hours of delay within the cordon each day, and he might easily cause another 3 hours of delay outside of the cordon.

    That person might choose to commute by car because it saves him 1 hour per day compared with taking the Long Island Railroad – but he doesn’t consider that it costs other people 8 hours of delay.

    It is amazing how much the costs of that decision outweigh the benefits, even if we ignore all the environmental costs and just take account of the time saved and wasted. If someone decides to commute to Manhattan by car, the time wasted is about 8 times as much as the time saved.

  • Brian

    What was the baseline level of traffic before the additional vehicle entered the network? I think it’s important to know this in order to interpret the results. For example in the inbound AM peak, if the traffic in the network amounts to 1000 vehicles, a marginal increase in delay of 2 hours over those 1,000 vehicles is 2hrs/1000 = 7.2 seconds/veh. If the number of vehicles in 10,000 is 0.72 seconds/vehicle.

  • Charles Siegel (#8): The delay costs in our Time Thieves paper are calculated outside as well as inside the cordon. In fact, for our typical trip — 10 or 11 miles to get to the CBD, 1-1.5 miles within; and then the same in reverse for the outbound leg — the outside delays exceed the inside delays.

    Brian (#9): The baseline levels of traffic differ hugely with the time period of the trip.

    Both you guys: I’m glad you’re taking my delay-modeling seriously. Please download the spreadsheet model (link: here) and go into the “Delays” worksheet. The info you’re asking about should be there.

  • maaaty

    Charles K.,

    This is staggering work. The question has occurred to me but I never have grasped how to measure it. And your framing and historical context are nothing short of genius: “the first century of urban traffic was one futile effort after another to stop time theft.” I look forward to hearing more from you.


Congestion Pricing Q&A With Rohit Aggarwala, Part 1

Too many unanswered questions. Among New York State Assembly Democrats, that has been one of the most frequent criticisms of Mayor Bloomberg’s proposal for a three-year congestion pricing pilot project in New York City. Last month, Lower Manhattan Assembly member Deborah Glick said that she and her colleagues were “confronted with a dearth of information […]

What If We Measure Streets for Walking the Way We Measure Streets for Cars?

“What you measure is what you get,” the saying goes. In transportation, the dominant metrics are all about moving motor vehicle traffic, so America has built a transportation network that moves a lot of cars. Our streets may be dangerous, expensive, and inefficient, but they do process huge volumes of motor vehicles. A quintessentially American transportation metric — […]

Queens Pricing Opponent Is Right: $8 Is Crazy

Drivers who take an East River bridge would have to pay the $8 congestion fee when they reach Manhattan, even if they’re just passing through on their way to somewhere else. "That’s crazy," said City Councilman Leroy Comrie (D-Queens), who voted against it Monday. "That’s one of the reasons I’m so adamant against the plan. […]

Another Tall Tale About Congestion From the Texas Transportation Institute

Crossposted from City Observatory. Everything is bigger in Texas — which must be why, for the past 30 years, the Texas Transportation Institute (TTI) has basically cornered the market for telling whoppers about the supposed toll that traffic congestion takes on the nation’s economy. Today, they’re back with a new report, “The Urban Mobility Scorecard,” which purports to […]