Upton Sinclair said “It is difficult to get a man to understand something when his salary depends on his not understanding it.” Here are official USDOT vehicle forecasts compared to actual traffic data.


Note that only March 2001 to November 2001 and Dec 2007 to June 2009 were recessions. More over at Sightline, who provided the graph.

This is actually important, because we wouldn’t think we need huge highway mega-projects if it weren’t for these forecasts.

38 Replies to “Traffic Forecast v Reality”

  1. Does anyone know of studies into the fundamental causes behind the recent VMT turnover?

    For example, if it’s primarily due to an increase in telecommuting with the rise of the internet, we might expect this turnover to hit some kind of floor before trending upward again.

    If, on the other hand, it’s due to some other fundamental cultural or economic shift (young people hate driving, gas is more expensive, etc.), the downward trend may continue.

    1. It appears to be due to several things at once. Young people *do* dislike driving, though — you can’t use your cellphone while driving — so that’s probably the most important, and that one’s permanent.

      1. You shouldn’t use it, but oh so many people do. I wouldn’t be surprised if we start seeing increases in road fatalities due to distracted driving, after seeing the steep declines of the past few decades.

    2. Fuel is probably just as important. It’s risen in price much faster than the rest of CPI (in fact, it’s one of the components of CPI that pulls the whole up so much).

      1. If fuel’s a big part of it, then the adoption of electric cars will push VMT back up. Although the sticker prices are higher so a bunch of people will still go carless.

        I think cellphones, fuel costs, and the impoverishment of everyone except the 1% are all part of it.

        The impoverishment of the 99% has been going on since 1980 and shows no signs of stopping soon.

  2. It would be interesting to see the data from our state, and even the Puget Sound region, rather than national data, in order to draw any meaningful conclusions locally.

  3. Careful which rocks we look under.
    Light Rail ridership was based a lot on time saved because, you know,’traffic is getting a lot worse each year’, as all the graphs showed at the time. Look at the Sound Move documentation. Hours saved by building rail is what the FTA wanted to see for justification, as well as several other factors, so that’s what they got.
    I can’t recall when they came to there senses, and downplayed that in the travel demand models, but the nasty graph above has driven a lot of spending in the last two decades both on the highway and HCT side.

    1. I suppose it depends how those that switched out of driving get to work now. Do you know what increases ridership more than congestion? Ridership.

      Actually, don’t let the word “traffic” in the article confuse you – they’re talking about traffic volumes. that’s quite a bit different from “traffic” in the word’s common usage. Congestion can decrease traffic volumes, especially if it’s peak-only. And road building can increase traffic volumes.

      1. Probably, sure. Depending on where the population is growing, and how much excess capacity exists on highways. Certainly if we’re talking about an increase in population in the suburbs the (traffic volumes) + (ridership) will scale linearly, though it’s probably not easy to tell which term rises faster without knowing the details. But congestion doesn’t scale linearly. It’s a good guess that it will generally rise with population though.

    2. “Hours saved” (along with the “congestion reduction” shibboleth) is a terrible way to judge a transit project. You can’t change congestion, except at artificial bottlenecks. People will tolerate a certain amount of it and grow into uncongested corridors until they are somewhat congested.

      “Capacity added,” both in terms of transportation capacity and growth/development capacity, makes a whole lot more sense.

      1. We agree. Actually, now that light rail is being built, but will not likely fill to capacity for many decades, is reason enough to tell the DOT, thank you very much for all the Mega Plans, but rail negates the need for more freeway lanes, except at some choke points – sort of Vancouver BC, minus all the lanes in place now.
        “See all the money we’ve saved you”
        A thank you would be nice!

  4. It would help if the y-axis were labelled, either here or on Sightline. As it is, about all this tells us is that forecasts of whatever is being graphed haven’t been good.

      1. Thanks.

        Seems to me that just because aggregate VMT is down, doesn’t necessarily mean that the megaprojects don’t make sense.

        For example, if this is just a reflection of fewer long distance leisure trips, and a move from rural areas and exurbs to suburbs, it’s possible that there is still a need for more peak capacity in cities.

        In short, the DOT sucks at projection, and as a result we need to be carful about taking their forecasts at face value, but that isn’t much of an argument that a particular megaproject is unjustified.

      2. Sure, look at individual projects. My favorite is the 520 project. State DOTs consistently have this terrible prediction strategy (or the curves above would be curvy), and VMT is consistently flattening or dropping for most roads.

      3. Yes, but when you are spending billions $, the burden is on you to make sure you need it. When you ignore the facts, as in the 520 case, it seems like maybe you aren’t making the right decisions.

        It seems fairly obvious we don’t need a larger bridge.

      4. @Andrew: The things that are causing the new 520 bridge to be larger than the current one are HOV lanes, the bike path, and wider shoulders. How would you cut the size of the new bridge?

        – Take GP capacity to one lane in each direction? You can sell this to me easily enough, but probably not to most voters and politicians.
        – Remove the HOV lanes? Actually they usually aren’t necessary on the floating bridge, and with good interchange design on the west end and HOV lanes all the way up to the edge of the bridge from the east it might be possible to eliminate them without hurting bus times too much. But if that leads to erosion of HOV facilities through the real bottlenecks it’s a big mistake.
        – Trim the shoulders down? Enough to make a big dent in the bridge’s cost?
        – Remove the bike path? Over my dead body.

        I will grant this: the ramp capacity is ridiculous, and increasing exit capacity to Montlake is just stupidly bad engineering.

      5. Al, I think you got it. I think take the HOV off, and keep the tolls, and you’re good. Leave the bike line (which is like half a car lane in one direction).

        Definitely, scale down the massive interchange on the montlake side.

      6. William, last I checked, *rural* VMT was actually up. It’s city and suburb VMT which is down.

  5. Highway planners in general have a problem imagining what happens when the space runs out that complete automobile-dependence requires. Transit planners’ weakness is the length of time their projects take, and what to do in the meantime.

    To me, highways and arterial streets will be better for automobile travel in the short run if they’re designed for eventual conversion to transit when time comes. Two-way transit lanes between Northgate and Convention Place would not have made I-5 a worse freeway- something that can still be corrected.

    And I personally sat at the table thirty years ago, when the world’s top rail engineers designed the Downtown Seattle Transit Tunnel with capacity that hasn’t been used in over twenty years of operations. However long North-Link takes to complete, Lynnwood and Everett service need not be stuck in Seattle traffic now.

    Granted, expense and work in the present, like adding staging supervisors in the Downtown Tunnel, training drivers from other systems to handle joint operations, and buying dual-power buses, means harder work for the system right now.

    But whatever happens over two or three more decades, transportation will be, and will have been, enough better to justify our trouble. Dreams of the future are bad excuses for avoiding action in the present.

    Mark Dublin

  6. “This is actually important, because we wouldn’t think we need huge highway mega-projects if it weren’t for these forecasts.”

    A couple thoughts on this statement:

    1. Safety is the single greatest reason for replacing both SR 520 and the Alaskan Way Viaduct. Both assets are at the end of their useful lives.
    2. If the reason these mega-projects are being constructed is to satisfy future demand, why is WSDOT reducing capacity on the SR 99 mainline (existing AWV relative to a four-lane tunnel)? Further, all funding constraints aside, if WSDOT really wanted to increase throughput on the new facilities, why is it managing demand with tolls?

    It seems a little misguided to claim that the only “need” for these mega-projects is to meet projections for future demand.

    1. The AWV project is maintaining the current capacity of SR 99 for trips that stay on 99 through downtown. The Battery Street Tunnel has 2 lanes each way and so will the new tunnel.

    2. 1. Safety is the single greatest reason for replacing both SR 520 and the Alaskan Way Viaduct. Both assets are at the end of their useful lives.
      2. If the reason these mega-projects are being constructed is to satisfy future demand, why is WSDOT reducing capacity on the SR 99 mainline (existing AWV relative to a four-lane tunnel)? Further, all funding constraints aside, if WSDOT really wanted to increase throughput on the new facilities, why is it managing demand with tolls?

      It seems a little misguided to claim that the only “need” for these mega-projects is to meet projections for future demand

      First, I agree sort of about the SR tunnel.
      Second, 520 doesn’t need to be bigger

      Third, the largest, most expensive mega-project (I-405 widenning) would almost certainly be better served by simply congestion charging. No $15 billion needed. Plus, they would be able to close the funding gap on the other projects they need money for (520, SR-99 tunnel), win, win, win.

      1. Yeah, but inflation can’t be accounting for the increase. When the 2001 study was completed, that had $4.5 Billion for the GP lanes, and about another $1.5 Billion for ancillary work. (Budget dollars)

        Then, a few years ago, WDOT had a $10 Billion figure on their website, also supposedly in Budget dollars, although a few years later than the 2001 study.

        If the $15 Billion is an actual WSDOT figure, I’d start to wonder how many other projects are being rolled into this.

        Hey, maybe I can make the case for getting my driveway redone… Hmmmm.

    3. Due to excluding traffic to/from Downtown, the waterfront, and Interbay/Ballard from the DBT, capacity (and speed) for traffic skipping downtown on the main line of 99 will certainly increase, though the utility of the road decreases due to this.

      It’s absolutely true that the viaduct needed replacing, and that non-transportation concerns featured heavily in the choice of such an expensive and risky replacement. Of course, traffic projections have something to do with the belief that two separate freeways are needed across downtown, and have something to do with the design of the surface road on Alaskan Way. But the tunnel is such a lousy project that it doesn’t even make sense from the typical auto-centric WSDOT perspective.

  7. I think there have been a few major trends over the years in the U. S. A.:

    1865-1900 — People moved to the city to live and work
    1945-1980 — Middle and upper class families moved to the suburbs to live
    1980-2000 — Companies moved to the suburbs
    1995-Present — Young middle and upper class people move back to the city; many create families
    2005-Present – Companies move back to the city, or to existing urban centers in the suburbs

    I think we are seeing that last trend play out. There is still plenty of “suburban” office growth, but it is occurring in areas like downtown Bellevue (which, if you didn’t know any better, you would assume was simply a small city). Drive by suburban office parks and the growth has pretty much stalled (in areas like Eastgate and Factoria). These are spots that looked like they could become the next downtown Bellevue a few years ago, but instead we are seeing growth in South Lake Union, or just about everywhere within a half hour walk of downtown.

    Meanwhile, the public transportation system for the suburban office parks are catching up. There are plenty of buses to downtown Bellevue — and those bus rides are a lot faster the more they add carpool lanes.

    At the same time, cities are as popular as ever. All those cranes between downtown and the ship canal are building office buildings or apartments. It is hard to tell which. From a transportation standpoint, either one will lead to a lot less traffic.

  8. The travel demand models that produce these forecasts are dependent on the projections of population and employment growth that come from MPOs (PSRC here in Seattle). It would be interesting (and more insightful) to look at how well those projections have been lining up with reality.

    1. Either way, the model is overly simplistic since it results in a single value for all time: a constant growth rate. That they’re not even dynamical is a bit ridiculous. Although not completely apples-to-apple, 15 years ago in my undergraduate PDE class we solved traffic flow models that were clearly more sophisticated than what is being used for the 520 analysis. So I’m absolutely certain there’s more sophisticated state-of-the-art modeling out there that could/should be used.

  9. If you check out the Historical Monthly VMT Report on the FHWA website, and look to see the trend, it’s not so far off that the highway engineers have not changed the slope of the forcast based on the recent trends.

    The trendline they might be basing their analysis on could be because they used data from the late 1970’s/early 1980’s as the start point.

    The problem with the above graphic is that we don’t know where the data points come from, and then there’s the old game of having different scaling of the x and y axes, generating different looking results, and conclusions.

    The other problem with the above graphic is that it doesn’t match the long-range planning horizon that is used.

    If the cost/benefit analysis is based on a 30/40 year duration it does make a difference.

  10. There is a lot of pressure on forecasters to show higher numbers, so don’t blame them. Showing flat volumes in many studies aren’t growing can get a forecaster fired — or at least taken off the assignment! When there are other professionals (such as other paid professional planners and engineers) whose job depends on building, they attack anyone who doesn’t validate their bias with a vengeance — even to the point of trying to ruin their careers.

  11. The graph says it all. It adds further credence to supporting a “maintenance first, bottleneck relief second” priority for highway funding, with low priority for adding anything new (which we aren’t paying to maintain as it is, for we’re not keeping up with existing road maintenance).

Comments are closed.