How Induced Travel Causes New Delay
The Example of I-270 in Colorado
Six and a half miles of urban freeway, choked with cars and heavy trucks, blasting through a neighborhood with some of the worst air pollution in the entire United States. If the Colorado Dept. of Transportation has its way, I-270 will soon be widened from four lanes to six. State officials are citing modeling which says that widening I-270 would alleviate traffic congestion, reducing emissions and helping people get around more easily. It sounds great – except that this modeling didn’t take into account induced travel.
Ives Street recently conducted an independent technical analysis of the possible I-270 widening. We included induced travel, and our findings are a little less rosy.
“Induced travel” is no mere hypothesis. It’s the proven phenomenon that expanding highway capacity causes more driving, and it’s been documented all around the world. That induced travel is a “fundamental law of road congestion” is not a controversial claim, and understanding of this principle is not confined to journals or conference proceedings. And yet too many of the tools used by transportation planners to forecast traffic and model transportation impacts fail to account for induced travel, leading planners to overestimate the benefits of roadway expansions.
In a recent collaboration with Earthjustice and GreenLatinos, Ives Street conducted an independent technical analysis of potential impacts of a proposed highway widening with a toll lane on I-270 in the Denver region. The modeling study conducted by the Colorado Department of Transportation (CDOT), like too many other studies of proposed widenings around the world, omitted induced travel altogether. Our analysis, on the other hand, took induced travel seriously. We estimated the total amount of travel that would be caused by a widening, we identified the most likely roads that these new trips would take, and we calculated the probable impacts of all those new trips on congestion and delay.
What we found won’t be a surprise to anyone who’s lived through a highway widening in their city: if widening I-270 gets rid of a bottleneck in one place, it will just create a new one somewhere else. For every minute a driver saves in faster traffic flow on I-270, another driver will lose a minute stuck in even worse traffic on I-25 or I-70, caused by all those new travelers enjoying faster speeds on the widened road.
This is a story about I-270 in Denver, but it’s also a story about so many other cities, and so many other roads, around the world. For most of a century, roadbuilding has been justified by modeling that optimistically overlooked the phenomenon of induced travel. But it’s not hard to include induced travel in an analysis.
Our study only needed data and tools that most planners already have at their disposal. We didn’t need fancy microsimulation software or integrated land-use models, we didn’t need a detailed synthetic population, we didn’t need a team of data scientists or hundreds of thousands of dollars. All we needed was a representation of the road network with estimated traffic speeds and volumes, some basic census data, and a willingness to take induced travel seriously.
What we did in Denver
We structured our analysis of traffic impacts from a possible I-270 widening around four methodological steps:
Representing the regional roadway network under current conditions
Estimating total induced vehicle-miles traveled (VMT) from the widening, using an industry-standard tool
Assigning new traffic to the roadway network
Predicting consequent changes in traffic congestion, delay, and access to jobs
These methods are described in full detail in our report, so I’ll only summarize them here.
First, representing the regional roadway network: Our analysis assumes no ambient growth in VMT. Historically, vehicle travel and economic growth have been closely linked, but that relationship has weakened substantially since the turn of the century. Many regions have seen population and GDP increase while per-capita driving has stabilized or declined. Growth in VMT is the result of government decisions to expand roadways, not an independent causal factor. The era of ‘predict and provide’ is over. And so, rather than assuming an increase in VMT over time, in our study we calibrated baseline conditions using observed speeds and volumes rather than relying on modeled relationships. We used high-resolution telemetry data from TomTom, but other planners could use other sources to represent the regional roadway network as it exists and functions today.
Second, we estimated induced travel from the I-270 widening using a literature-based, exogenous tool: the State Highway Induced Frequency of Travel (SHIFT) calculator from the Rocky Mountain Institute. SHIFT translates added lane-miles into an expected range of new vehicle travel based on empirical elasticities derived from decades of observed highway expansions. This step answers a simple but essential question: how much additional annual vehicle travel is likely to occur because capacity was added?
Third, we converted that estimate of total annual induced VMT into an estimate of average weekday peak-hour induced VMT. Then we assigned that travel across the regional network rather than confined to the expanded facility itself. Induced trips are full door-to-door journeys, and their impacts propagate onto connecting highways, arterials, and local streets. Assigning induced VMT network-wide makes it possible to evaluate where congestion and delay are likely to appear.
Finally, we used industry-standard “volume-over-capacity” congestion functions to translate changes in volume into changes in speed and delay. These functions are familiar to practitioners and do not introduce new theoretical complexity. There are more nuanced ways of accounting for the relationship between congestion and delay, but those would have only provided more detail, not a fundamentally different set of results. (We made sure to avoid dramatic over-capacity assignments, a common but controversial practice in which models are permitted to allocate more cars to a stretch of highway than it can physically hold).
The impacts of widening I-270
We found that adding new toll lanes to I-270 would have no net benefit to regional congestion. It might alleviate congestion on I-270, but it would just make traffic worse on all the roads that connect to I-270, especially I-25.
Once induced travel is assigned across the full network, the apparent benefit of the widening largely disappears. The additional traffic generated by faster conditions on I-270 flows primarily onto the two interstate corridors that connect directly to it. In our analysis, new congestion emerges most clearly on southbound I-25 in the segment approaching the I-270 interchange and on eastbound I-70 between the I-25 interchange and points farther east. These are already heavily trafficked highways during the morning peak hour, and even modest increases in volume in these congested roads will result in disproportionate increases in delay.
This figure from the report illustrates reduced delay on I-270 in shades of cyan/blue/black and increased delay on other roadways in yellow/orange/red. The colors indicate “change in vehicle-minutes of delay per mile, A.M. peak hour” – a value of +10,000, for example, indicates that along a mile-long stretch of highway, the equivalent of 10,000 drivers will each spend an extra minute in traffic relative to current conditions – more likely, something like 5,000 drivers will each spend an extra two minutes, or 2,500 drivers will each spend an extra 4 minutes, etc.
This finding – that widening I-270 would redistribute, rather than reduce, congestion – emerges even under conservative assumptions. Our analysis does not include time-of-day shifts, which would likely lead even more drivers to travel during the peak hour once new capacity is added. Including that effect would likely increase total peak-period traffic and worsen congestion outcomes in the widening scenario.
In this analysis, we demonstrate that it is not only possible but eminently feasible to include high-resolution analysis of induced congestion and delay in any analysis of a roadway expansion. When traffic modelers (not just CDOT!) leave induced travel out of their models, they’re leaving out half of the story. They’re going against “a substantial body of high-quality evidence.” The public deserves better.
More detail about our results, including an analysis of resulting access to jobs, is available in the full report. It also describes another, alternative scenario – the community-driven No-Widening Scenario in which the entirety of I-270 is converted to an express toll road.
The bigger picture
We have all seen the effects of induced demand play out in real cities over time. Freeways are widened, congestion initially improves, and then traffic returns, prompting calls for yet more capacity. The result is a familiar cycle: more lanes, more driving, and stagnant performance at the regional scale.
Many transportation planners and modelers, including many state departments of transportation and metropolitan planning organizations, are actively working to include induced travel in their planning and funding decisions. New guidance, climate legislation, and internal methodological reforms reflect a growing recognition that induced travel must be accounted for explicitly. Regulations like California’s CEQA require planners to account for induced demand; tools like the Rocky Mountain Institute’s SHIFT calculator make it easy for anyone to get a sense of the vehicle-miles traveled (VMT) that will be induced by a given capacity expansion. At the same time, however, many agencies and consultants still rely on conventional 20th-century modeling techniques which ignore induced travel altogether.
In many studies, induced demand is said to be “captured indirectly.” This may be through background growth assumptions that inflate future volumes regardless of whether or not a project is built, or through redistribution-only models (such as CDOT’s) that allow traffic to reroute but not increase in total. These approaches can produce internally consistent results, but they fail to represent net new travel caused by capacity expansion.
In other studies, such as CDOT’s modeling of I-270, planners acknowledge the possibility of induced travel but argue that it does not apply to their project because “the region is already built-out” or “surrounding land uses are inelastic to new capacity.” In the face of decades of evidence that induced travel is an ironclad law of transportation, the burden of proof should be on any planner who claims that their project is so unique that it is immune to this law.
Accounting for traffic delay caused by induced travel on a roadway network can and should be standard part of transportation planning practice. As we’ve shown in this study of I-270, there’s no technical reason not to include it. No city, region, or state government should accept the work of a traffic planning consultant who does not account for induced travel in their traffic forecasts. And no public citizen body should accept transportation planning decisions made by a government that does not account for induced travel.
What’s next for Ives Street
We’re looking forward to helping more cities around the country – and around the world – explicitly include induced demand in their transportation analysis and planning. The techniques we used for I-270 are applicable anywhere else, to road widenings of all kinds. If you’re working on a project where this kind of approach could be useful, please get in touch.
We’re also excited to continue developing the technology that we used in this analysis. Specifically, we’re looking into more sophisticated traffic-assignment functions, which would be capable of identifying the second-order implications of induced travel and which would be able to provide more precise information about where we might expect new bottlenecks to form.
And of course, we’re always continuing to develop our work on general access-to-destinations in more detail. You can look forward to more information about that effort in upcoming posts on this blog, Understanding Transport, here on Substack. Subscribe to stay in the loop.