Road-building advocates often point to studies that claim to show a multi-billion dollar economic cost due to lost time because of traffic congestion. These studies purport to elevate the ongestion problem from personal inconvenience to crippling economic burden. Despite frequent and repeated claims that traffic congestion somehow costs our economy billions, the methodologies used to make these claims don’t stand up to close scrutiny, and recent data show that driving and congestion are actually declining in the nation’s cities.
More deeply, the focus on trying to reduce congestion by increasing road capacity has proven to be a Sisyphean task because of the phenomenon of induced demand. Road widening triggers travelers to change travel patterns, take more and longer trips and in the long run leads to more sprawl—which further increases traffic. Because there’s no feasible way to build enough capacity to eliminate congestion, the supposed “costs” of congestion are a misleading illusion.
Inrix, a Seattle-based provider of traffic data is extremely clever and helpful: they’ve devised a way use data from commercial vehicle GPS systems and cell phones to estimate, minute-by-minute, how quickly traffic is moving on hundreds of thousands of miles of roads in the US and worldwide. Their technology enables us to have color-coded maps showing traffic levels on the roadway system that we can view in real time on our mobile devices. ?It’s a real boon to travel to know when and where you’ll confront delays.
Whether you’re a transportation researcher and planner, or just a traveling citizen, there are good reasons to be grateful to Inrix for using technology so creatively, and being public spirited enough to also open their archive of data to wide public access. We now have accurate, detailed, essentially real time data on the use of our transportation system that we simply didn’t have before. Potentially this kind of detailed data can be useful to transportation planners for evaluating trends how investments and policy changes affect travel behavior.
(If you go here you can check out Inrix’s traffic maps like these:)
Actually, no. Here’s the surprising thing: The Inrix data shows that nationwide traffic congestion is decreasing, not increasing. ?Inrix data show that the nationwide travel time index for the US has fallen from a level of 13.4 in February 2010 to 7.9 in January 2014.–a decline of 40 percent. The change in this index implies that the number of hours lost per month declined from 2 to 1.2 over the last four years.
The decline in traffic congestion hasn’t escaped the attention of Inrix. Two years ago, they issued a press release that proclaimed: “Traffic congestion plummets worldwide.”
The trend of declining traffic congestion is even grudgingly acknowledged—then completely ignored–in the? latest Inrix/Cebr report (page 16):
“We estimate that, between Q1 2011 and Q4 2013, the amount of wasted idling in congestion has seen an average annual decline of 8%.”
There’s a simple reason for the decline in traffic congestion: Americans are driving less. After increasing for decades, the amount of driving by the average American peaked and started declining in the middle of the last decade. Today, the typical American drives about seven percent less (25.6 miles per day compared to 27.6 miles per day in 2005).
The good news here is that the decline in auto travel is making major inroads into traffic congestion.
Not really. The Inrix/Cebr study predicts that over the next decade and a half, the average annual hours of delay experienced by a U.S. commuter will increase from about 22 hours per year to about 23.4 hours in 2030. (See page 40 of the Inrix/Cebr report). That works out to about 84 minutes of additional delay per year, and divided by a typical 200 days a year of work day commutes, represents an additional delay, compared to today’s levels of about 25 seconds per day of additional congestion related delay.
In recent years there has been a growing disconnect between engineers’ traffic projections—which predict ever increasing amounts of traffic and resulting congestion—and reality—declining levels of both. The State Smart Transportation Institute compiled a series of state traffic forecasts for the past 15 years and compared them to actual travel trends(shown in chart below). The forecasts repeatedly predicted robust annual growth averaging about 1.8 percent per year—and for the past decade have continually missed the mark. And even as actual traffic levels have declined, forecasters have continued to predict a robust return to growth.
The Inrix/Cebr report follows very much in this mold. It predicts (on page 37) that total passenger vehicle miles will increase about 19 percent in the United States between 2013 and 2030. This works out to an annual increase of about 1 percent per year. For the past decade (2004-2014) the annual rate of increase of total vehicle miles traveled has been 0.1 percent per year. Consequently, the Cebr forecast of future traffic and congestion relies on a rapid and sustained increase in vehicle travel—to rates we haven’t seen in the US in more than a decade. The implied Cebr growth path, compared to the historical trend in total vehicle travel is shown in the following chart.
A central pillar of the congestion cost argument is that every additional minute that a person spends traveling is associated with an economic cost. It’s not clear from the Inrix/Cebr report what value of time was used to estimate travel time “costs.” Other studies, like the Urban Mobility Report use high values of time—more than $15 per hour to value every additional minute in travel.
Practical evidence with variable tolling shows that motorists don’t value travel delays at anything approach the dollar amounts estimated in this report. Fewer than ten percent of motorists using a Seattle area highway were willing to pay even as much as $12.00 per hour of time saved to use road’s tolled express lanes. These results imply that the vast majority of people value their time savings at much less than $12.00 per hour.
According to Inrix, the “average” peak hour traveler experiences six or seven minutes of delay per day, or about three minutes per trip–and many of us experience much less than that. For most motorists additional travel time associated with congestion is often “infra-marginal”–too small to be noticed, given the daily variation in travel time. Most motorists are likely to be unaware of travel time differences of two or three minutes or less on the typical 20-minute commute. Travelers generally don’t attach a value to a difference they can’t perceive.
The way the travel time index is used to compute “costs” is to compare what are called “free flow” speeds with the actual speed on the roadway system.
The baseline assumption used by the travel time index–that any reduction in speed from free flow levels on the roadway system represents a “cost” is simply wrong. Free flow travel speeds on many roads exceed the posted speed limit, and so the travel time index methodology can compute a cost associated with the time lost to travelers who cannot exceed the speed limit. Todd Litman estimates that a significant fraction of what is labeled congestion costs is actually compliance with posted speed limits.
Somewhere between a quarter to half of estimated congestion costs represent speed compliance. ?In more congested cities like Los Angeles and Miami, a majority of all time losses attributed to congestion involve slowing down to the posted speed limit.
It’s also worth noting that while measured or theoretical free flow speeds are often at or above the posted limit, these are not the speeds that maximize the capacity of the roadway–maximum traffic capacity on freeways, for example, is generally estimated to occur at between 40 and 45 miles per hour. As cars go faster than this speed, which varies by roadway, the cars actually spread out further—increase their following distance—and as a result the road carries fewer cars per hour than it would carry at a slightly slower speed. Using a higher speed as the baseline for computing congestion costs implies that we ought to build enough freeway space not just to carry the maximum number of vehicles, but to have enough extra space that they can drive somewhat faster.
It’s far from clear how much money we would have to spend to create additional capacity to try to avoid or prevent the so-called costs of congestion. It’s highly likely that the annualized cost would be many times the congestion cost estimates presented here. As a result, there’s no evidence that the costs of congestion represent any net cost to society. The time theoretically “lost” to congestion only has a real value if we could build a transportation system that would eliminate congestion for less than the so-called cost of delays.
Totally eliminating congestion costs would require providing enough transportation system capacity that there would be no peak hour delays to any traveler. Because adding peak capacity—new freeway lanes—especially in dense cities is extremely expensive, it is likely the cost would exceed by an order of magnitude or more the value of congestion costs. It would be prohibitively expensive, for example, to build enough freeway lanes in Los Angeles to accommodate all the cars that would travel at the peak hour. It would actually be a much bigger “drain on the economy” if we tried to expand capacity to reduce supposed congestion “costs.”
The big problem with expanding capacity to try and reduce congestion is that it doesn’t work. The road system, like nature, abhors a vacuum. Additional road capacity typically prompts people to change their travel patterns (taking more frequent trips, re-routing, and changing departure times), a process called “induced demand.” Even if would could costlessly expand capacity, there is strong evidence that induced demand would cause roadways to become congested again. This process has been repeated in cities around the country: when new roads are built to reduce congestion, they lead other travelers to change the route, timing or destination of their trips in ways that lead to even more vehicle miles of travel.
One of the classic examples is a newly opened billion dollar carpool lane on the Los Angeles 405 freeway (ironically, the site of the 2011 “Carmaggedon” freeway closing). After re-opening with an additional travel lane in each direction, data gathered by Inrix showed that travel times were actually longer than they were prior to project construction.
This phenomenon is so well understood that two researchers—Gilles Duranton and Matt Turner have christened it “the fundamental law of road congestion.” Vehicle travel increases in direct proportion to increases in highway capacity and that as a result capacity expansion has no effect on the levels of traffic congestion. In short, newly expanded roadways typically fill up rapidly–and travel times don’t improve. This cycle is destined to repeat itself in dense urban areas when we don’t ask road users to pay–via tolls–the full costs associated with using the roadway.
While the travel time index is reasonably useful for comparing changes over time in a particular location, it is misleading, and often entirely wrong when used as a tool to compare different metropolitan areas, and to estimate the supposed “costs” of congestion. A key problem is that the travel time index ignores differences in trip length and density between cities. Its construction makes it look like sprawling metropolises with very long commutes have less congestion; it ignores the travel time penalty associated with sprawl. (Check out this report to find out more.)
The places in the country with the greatest time loss to travel are not those with the highest travel time indexes, but rather those with the lowest density and with the highest levels of sprawl. Because housing, jobs, and other destinations are so spread out, people travel more a lose more time traveling these longer distances (albeit at faster speeds) than their peers in more compact metro areas that may have higher levels of congestion.
Engineers have repeatedly predicted Carmaggedon, but the truth is, it never happens.
One of the most widely anticipated “Carmaggedons” was a planned weekend closure of the a ten-mile stretch of the 405 Freeway in Los Angeles, one of the nation’s busiest, in 2011. In the days leading up to the closure, experts warned of the “inevitable” and “epic” delays that would ripple throughout the region.
This story has been repeated time and again around the country. In Minneapolis, the tragic collapse of the I-35W bridge in 2007 severed a major transportation artery that carried more than 140,000 vehicles per day. But follow up studies by David Levinson and his colleagues at the University of Minnesota showed even with the bridge out of commission for nearly a year, region-wide traffic delays returned to normal within weeks as travelers adjusted their routes and departure times.
There’s a good reason why these Carmaggedon forecasts are almost never realized. Travelers aren’t mindless lemmings–in the face of congestion, they make prudent adjustments to their schedules and travel plans, minimizing the negative effects of congestion. Especially when there’s some advance warning, or over time, as travelers learn when and where they encounter delays—travel plans change. Crashes and other “non-recurring” congestion are impossible to predict, and therefore avoid—and crashes account for a large share of time lost to congestion. But the garden variety evening rush hour is well understood by most travelers. Traffic engineers repeated forecasts of horrendous congestion suggest that the letters “PE” may really stand for pathological exaggeration.
What careful studies have shown, in fact, is that throughout most recorded history, across countries, and regardless of the technology used, daily travel times tend to average about one hour per day. Often referred to as Marchetti’s constant, this observation implies that as travel speeds improve (due to better technology or less congestion) people travel more. This has been true even with older transportation technologies including walking and the streetcar. Building more roads or faster modes simply encourages people to spread out further as they look for jobs and housing.
New research from Alex Anas suggests that the reverse is true as well. His analysis shows that travel times have been relatively stable in US metropolitan areas. His modeling of the Chicago area transportation system suggests that as congestion increases due to population growth, households gradually take steps to reduce their commute distances and that this coupled with mode shifts and other travel pattern changes tend to keep travel times stable.
The travel budget theory suggests that within these limits (say a 20 to 30 minute one way journey to work), people place little value on spending (or saving) time out of their travel time budgets.
Basically what the travel time index does is treat as “losses” any extra time it takes to travel between any two points compared to perfect conditions. It’s difficult to argue that such differences between some imagined perfect state and actual performance rise to the level of actual economic costs.
It would be just as valid, for example, to argue that the failure to adopt supersonic air travel is imposing time losses on international travelers. The last operating supersonic Concorde jetliner was retired from revenue service in 2001. Travel times with the Concorde were about half as long as travel times with subsonic passenger jets–about 3.5 hours between New York and London, compared to about 7 hours. This is a travel time index of 2.0–actual travel takes twice as long as it might take under ideal conditions (if all international flights were via the Concorde). As a result of the lack of supersonic travel, the “travel time index” approach implies the 187 million annual US international passengers incur losses of about $11.7 billion dollars per year (the average 3,200 mile international flight takes 6 hours rather than 3, and time is valued at $20 per hour). Data on international air travel are from the US DOT.
It’s fair to argue, of course, this “cost” isn’t real, because it would be far more expensive to replace all international flights with supersonic transports than the value of any time saving would be worth. The added cost of the more expensive planes, and greater fuel consumption (and environmental damage) would outweigh value of the travel time savings, and in the real world no one would pay those costs. But that’s pretty much exactly the same as with the supposed cost associated with peak hour road congestion.
Fortunately, we at City Observatory have done the crucial work to answer this question.
Because an estimated 100 million of us American workers can’t begin a productive work day without an early morning jolt of caffeine, and because one-third of these coffee drinkers regularly consume espresso drinks, lattes and cappuccinos, there is significant and growing congestion in coffee lines around the country. That’s costing us a lot of money. Consider these facts:
Read more in our post “The Cappuccino Congestion Index,” which also features this groundbreaking chart:
The Victoria Transportation Policy Institute’s Todd Litman has prepared his own timely de-bunking of the new Inrix/Cebr report for Planetizen.
There have been several in-depth examinations of the travel time index and the claims made about the economic cost of congestion.
Our own report for CEOs for Cities, contains a detailed critique of the Urban Mobility Report, and flags flaws in its estimates of traffic congestion and fuel consumption. It also presents measures of the travel time costs associated with sprawl (which are concealed by travel time index calculations). Finally, it outlines alternative measures for better assessing urban transportation system performance and guiding policy and investment.
Todd Litman has also prepared a Congestion Costing Critique–a thorough deconstruction of the Urban Mobility Report, as well as engaging in a detailed debate with the report’s authors.? He questions everything from the reliability of the report’s speed estimates, value of time, and fuel consumption models to the lack of peer review and the author’s credulity about how the report is perceived and used.? This report also offers a lengthy summary of the published literature discussing travel time measurement and valuation.