Transpo Agencies Are Terrible at Predicting Traffic Levels

This chart contrasts state DOTs' projected traffic volumes with those actually recorded by the Federal Highway Administration. Image: ## SSTI##
Combined traffic projections from state and regional transportation agencies (the colored lines) have been wildly off the mark (the black line shows real traffic levels) for more than a decade. Image: ##

Americans’ travel behavior is changing dramatically. It seems like not a week passes without a new report about the decline in driving. But are state and local transportation agencies — which are responsible for much of the nation’s highway and transportation planning — keeping up with the facts on the ground? A review of the evidence by the State Smart Transportation Initiative finds the answer is a definitive “No.”

Forecasts and assumptions about ever-increasing traffic are often used to justify agency decisions to expand roads. But these assumptions are increasingly divorced from reality. In fact, state and regional agencies aren’t just wrong some of the time. State DOTs and metropolitan planning organizations are getting it wrong every year, over and over again, by significant margins, according to SSTI’s analysis.

In their most recent reporting to the Federal Highway Administration, state and regional transportation agencies used data from 2008 to predict that traffic volumes would reach a combined 3.3 trillion miles nationally in 2012. Last year, a few months after that forecast was publicly released, real-world data already showed that the forecast wasn’t even close. Transportation agencies had collectively overestimated how many miles Americans would drive in 2012 by 11 percent. That is the equivalent of adding five “average-sized” states to the total, SSTI reports.

What’s worse, these wildly incorrect traffic assumptions are routinely used to justify costly road expansions.

SSTI reviewed every 20-year traffic forecast submitted by state and regional agencies to FHWA since 1999 (these predictions are in a document called the Conditions and Performance Report to Congress). It turns out that the 20-year projections overestimated future traffic volumes in every single year the reports could be compared against data on actual miles driven by Americans. The 1999 report, for example, overestimated actual driving in 2012 by a whopping 22 percent.

SSTI’s Eric Sundquist concluded that states and MPOs “generally have not updated their models and assumptions to account for current conditions, as if they expect the year to be 1980 forever.”

26 thoughts on Transpo Agencies Are Terrible at Predicting Traffic Levels

  1. If these state agencies went with the “black line” reality instead of those colored-line projections, would the money now not needed for massive megahighways fund meaningful transit expansion sooner rather than later? Or is it too little too late?

  2. Not only are all the estimates far off, but they appear to be off almost in parallel to each other. So it’s not just incorrect, but also lazy since they are not correcting at all each year, except where the line crosses y.

    Also, why are the estimates flat line? Even just a basic graphing calculator can find a non linear trend. Just about any one who took college math (and didn’t fall asleep every day) could do a more accurate prediction. And these people call themselves engineers? Their school should lose its accreditation.

  3. It’s probably worse than that. They’re probably deliberately misleading people for financial reasons. Do you really think your average carhead state legislator cares or wants to understand anything more complex and nuanced than a straight line anyway?

  4. Since every slope has the same line, I’d say they are probably using the same model year after year. What’s fascinating is that it was basically an accurate prediction at first (from 1997 to 2000) and then becomes progressively worse as an estimate each subsequent year. They are obviously using models that were accurate for the 90’s but we’re now in the second decade of the new millennium.

  5. Kind of like transit agencies in the 1970s with ridership and cost predictions. A Desire Named Streetcar. Read it.

  6. The federal money is primarily being used for maintenance and managing existing roadway assets. The federal money is not primarily being used for massive new highway expansion projects. What transit needs to do (with WMATA as a prime example) is to focus on maintenance of existing assets before system expansion.

  7. State governments control a lot of money too. And states seem to have little problem getting billions in federal loans for massive road-only infrastructure projects, and then pitching in billions of their own too. But it is agreed that much of the money flying around is going toward upkeep. Not all of it is going toward new roadways and thus should be redirected to transit projects.

  8. How does this account for places like Northern Virginia, where roadway construction is still lagging far behind demand? So many people have no viable option but to spend every weekday morning and evening idling in gridlocked traffic for hours. Even if traffic doesn’t increase in the future, roads should be expanded and added simply to support the current need.

  9. Because the most economically efficient amount of road is not the amount where there’s never congestion, but the amount where the cost of that congestion (MC) equals the cost of adding another lane (MR; therefore, when MR=MC). Therefore, the presence of congestion doesn’t prove the need to add roads or lanes.

  10. Perhaps more explicitly – the absence of congestion proves the absence of need, but the presence of congestion only proves the presence of need when the congestion is so bad that residents would be willing to destroy their own homes or businesses or parks in order to avoid the congestion.

  11. There’s no reason to look for a non-linear trend. A linear trend fits a simple model of how traffic patterns develop. Once any factors other than population growth are relevant to the change in traffic patterns, these factors are likely to develop in such unpredictable ways that even a non-linear trend extrapolated from the past will be no better than the linear trend. At that point one needs models rather than extrapolation.

  12. Don’t forget the part about being willing to give themselves cancer by way of the lethal poisons put out by cars.

  13. The congestion indicates that the location is ready for an exclusive BRT lane, imho. See 36 Boulder to Denver. (I may have that example wrong)

  14. OK will they are behind the times for sure. But maybe they are just waiting for the current Great Recession to end. But I agree I think we are seeing a permanent change in the trend.

  15. How is the cost of congestion being measured, though? Is the health and well-being of commuters taken into consideration at all? What about the additional CO2 emissions, or all the productive time being wasted?

  16. The problem is that they are projecting, not predicting. A projection is just math; taking a historic trend and applying it to the future. Their projections are mathematically correct. The issue is that the historic trend they are using is faulty. This is done entirely intentionally in order to justify road projects. As long as “corporations are people, my friend!” this willful and intentional faulty math will continue.

  17. Road infrastructure is *hugely* subsidized in North America (and most other infrastructure too – this is what enables ‘suburban living’ to be so inexpensive). We do not have the money or political will-power to maintain our existing infrastructure (that is why our infrastructure is falling apart), and yet somehow we still manage to find the funds for additional road expansions. Well, the music has stopped, and anyone who is locked into a long-distance commute is going to be left without a chair to sit on.

    Would you be willing to pay tolls to have uncongested roads? If you are not willing to pay the true price of your travel decisions, then the market will use congestion to resolve the imbalance between supply and demand. And why should we expand roads which have congestion for two to four hours a day? What about the rest of the time when there is not any congestion? Who is going to pay for that?

  18. I’m only speaking for Northern Virginia, but some of the highways out there like 395 and 495 are congested almost any time of day. After the evening rush, around 9 or 10 pm, the roadwork starts so that keeps the traffic moving sluggishly throughout the night. I’m pretty sure the money is there, given all the high-income residents paying taxes, but it’s controlled by the folks in Richmond who are out of touch with what the northern part of the state needs (including more/better public transit in addition to more roads).

    I no longer live in Virginia, but I would have been willing to pay tolls. I come from a state where they’re a normal part of using the highways so I’m used to them. And while I prefer public transit, my driving commute still would have been significantly cheaper even with tolls factored in.

    When my mother-in-law bought her house in the 70’s it took 20 minutes to drive into the city. Now it’s more like 1-2 hours. That’s a big quality of life decrease and it’s not that easy for everyone to just pick up and move to a more urban area.

  19. DC is probably one of the more complicated jurisdictional areas in the country! Developments in one state can cause congestion in another, and DC itself has very limited power and even less control over regional development. If you are interested in this stuff, lots of info can be found on the Metropolitan Washington Council of Governments site:

Leave a Reply

Your email address will not be published. Required fields are marked *


Driving Declines Spell Big Trouble for Turnpikes

What the New Jersey Turnpike Authority did in 2005 was no different than what almost every other state and regional transportation agency was doing at the time. It predicted that traffic volumes would rise at a healthy clip every year for about 30 years into the future. Then it estimated its revenues based on those […]

Court: Don’t Spend Billions on Outdated Travel Forecasts

Cross-posted from City Observatory.  Last week, the Washington Post reported that the U.S. District Court in Washington, D.C., has ordered new ridership projections for the proposed Purple Line light rail line, which will connect a series of Maryland suburbs. Like any multi-billion dollar project that serves a densely settled metropolitan area—and this one connects some […]

Southern California Road Agency Courts Bankruptcy With Highway Addition

Today, U.S. PIRG and the Frontier Group released a new report, “Highway Boondoggles: Wasted Money and America’s Transportation Future.” In it, they examine 11 of the most wasteful, least justifiable road projects underway in America right now. This week we’ve previewed the report with posts about the proposed Effingham Parkway in Savannah, Georgia and the harebrained scheme to […]