daniel herriges is an urban planner and writer for the online journal Strong Towns, a nonprofit advocating fiscally-sustainable solutions to urban problems.
Illustrations by Roger Chouinard.
Published October 20, 2017
Few things may be certain in this world, but in the modern American cosmology, “traffic” occupies a spot right up there with death and taxes. It scarcely matters where you live — a city of 60,000 or 600,000, amid old Boston brownstones or in a sweltering Sun Belt suburb.
Indeed, “the traffic is awful and getting worse” is such a truism of American life that articles on the subject seem to exist outside of time. A search of the archives of my local newspaper turned up hundreds of mentions of the phrase “traffic congestion” over the past couple of decades. Stories from 2002 are effectively interchangeable with ones from last week. It seems we’ve made about as much progress in conquering this scourge as we have in eradicating the common cold.
But what are we supposed to do about traffic congestion that we’re not already doing? Calls to put more money into infrastructure are ubiquitous (and a rare point of bipartisan consensus). But it’s far from clear that we’d know how to spend that money in ways that produce results if it were available.
Historically the default fix has been brute force: more roads, more lanes, bridges, tunnels, bypasses. But such projects in already built-up areas are numbingly expensive, and seem particularly misplaced at a time when most localities face daunting deferred road maintenance obligations with no plan to pay for them.
In any event, the benefits of expanding the system can be elusive. In Houston, a $2.8 billion expansion of the Katy Freeway (Interstate 10) brought it to a jaw-dropping 23 lanes at its widest segment. Yet, after some temporary traffic relief following the project’s completion, rush-hour travel times from downtown to suburban Pin Oak increased by 51 percent from 2011 to 2014. The five-year, $1.1 billion widening of Interstate 405 in Los Angeles has similarly been deemed a failure by local media: a study by traffic analytics firm Inrix found that, soon after its completion, afternoon speeds on the 405 were slower than before.
But if at first you don’t succeed … downtown Seattle’s Alaskan Way Viaduct replacement tunnel, plagued by more construction problems than you could imagine, is up to an estimated price tag of $3.2 billion. The eastern span of the San Francisco Bay Bridge was replaced for double that sum.
Not liking where this money hemorrhage seems to lead, another camp responds to congestion by calling not for more roads but for fewer buildings, blaming rush-hour gridlock on overdevelopment. This narrative holds that runaway growth clogs city streets with new people in new cars, largely to the benefit of a cabal of real-estate moguls and business interests — the coalition that sociologist Harvey Molotch famously dubbed the “Growth Machine” in the 1970s. If we could put sensible limits on growth, claim these voices, traffic woes would dissipate.
This narrative holds much sway in cities such as Sarasota, Florida, a popular baby boomer retirement destination and the heart of the U.S.’s 11th fastest-growing metro area. Downtown Sarasota is dotted with construction cranes, jokingly referred to by Floridians as the state bird. There, a city election was recently won by a candidate affiliated with a growth-management advocacy group named STOP! (The name, complete with exclamation point, is not an acronym for anything.) The group and the candidate rose to local prominence over concern that development in Sarasota was outpacing the city’s capacity to plan for it, and that traffic congestion and other woes would result. Similar sentiments can be heard in cities and suburbs across America.
The overdevelopment story, however, is also problematic in a number of respects. Constraints on development can limit housing construction, making what housing there is dauntingly expensive. The National Low Income Housing Coalition estimates that, of the 3,007 counties in America, a full-time worker earning the minimum wage can afford a one-bedroom rental home in only 12 of them.
In any case, development restrictions may fail to achieve their aims. Without regional cooperation, one locality’s efforts to curb growth may not so much curb it as drive it to the nearest jurisdiction where regulation’s touch is lighter. The result is that most growth occurs on the suburban fringe, where public resistance to new construction is often weakest. These are places where commutes are long and amenities are fewer.
The traffic generated by population growth is not only a function of how many residents are added, but of how much, on average, they drive. For this reason, scattered suburban growth makes regional traffic worse than does high-intensity development in an urban center, where new residents would likely have driven less often and over shorter distances.
Responding rationally to traffic concerns requires that we better define the problem. Is traffic really getting worse? How so? And what measures stand the best shot of actually alleviating traffic woes at reasonable expense?
What a Traffic Problem Isn’t
The more you try to quantify the traffic problem, the more elusive it seems. There is scant evidence that it takes much longer to get places than it used to, and considerable evidence that costly highway expansions do little to shorten commutes or reduce congestion in the long run.
Yet the frequency with which American drivers complain about traffic is evidence of genuine, widespread frustration. Most adults depend on cars as their primary mode of transportation. So it makes sense to examine what a “traffic problem” actually consists of from the driver’s perspective. In broad strokes, it could mean one of three things:
- It takes me too long to get places I want to go.
- The time it takes to get places I would like to go is too variable or unpredictable.
- The experience of getting places I would like to go is unpleasant due to traffic.
The default assumption for many is that we have a traffic volume problem — as in “there are too many cars on the road.” Or, in a variation on the theme, we have a traffic delay problem: “All the other cars on the road are slowing me down.” This assumption is pervasive in public discourse, fed by publications like the Urban Mobility Report from the construction-industry-friendly Texas Transportation Institute and the Inrix congestion index. These simple explanations for what ails suggest a simple set of solutions. If the problem boils down to too many cars and not enough capacity, either we make more room for more cars or somehow make all those cars go somewhere else.
The paradox of the “too many cars” notion is that vast stretches of American automobile infrastructure sit underutilized most of the time. Even at afternoon rush hour, congestion is largely confined to a handful of busy, limited-access highways and arterial roads. We thus apparently have a traffic distribution problem, not a volume problem.
Hypothesis one above — that travel times are too long — is largely a misunderstanding born of repetition. In fact, travel times have been remarkably stable since they were first tracked nationally. The U.S. Census Bureau has asked workers about their commute time to work since the 1980 decennial census, and annually since 2005 through its American Community Survey. The long-term trend for the nation as a whole: a very modest increase over decades.
Despite much hype about the rise of “mega-commuters,” only about 8.5 percent of U.S. commuters travel more than 60 minutes one way to work. And public transit commuters are disproportionately represented in this number: among solo drivers, the share of these very long commutes falls to 6.5 percent.
Nor do we seem to be spending significantly more time driving for non-work travel. Since 2003, according to the Bureau of Labor Statistics, total time spent traveling per person per weekday has actually ticked down.
But while we’re not traveling for longer periods, we are traveling farther than previous generations did. The increase in miles traveled by the average person reflects the suburbanization of America and the increasing number of people who live in far-flung, low-density areas. And it does come with costs: far more infrastructure to maintain per capita, more air pollution and carbon emissions, more loss of farmland and wilderness. But more time spent sitting in traffic is not one of those costs. In fact, there is a lot of reason to doubt that we could significantly affect travel times by building large quantities of transportation infrastructure. The reason is a fascinating bit of theory embodied in a number dubbed “Marchetti’s constant.”
In 1994, the Italian physicist Cesare Marchetti proposed the idea of a bound on city size: a city would only grow to the extent that residents using the prevailing transportation technology of the era could accomplish their day’s travel in about an hour round-trip. Marchetti’s constant is remarkably accurate in explaining the upper bound on the size of ancient Greek and medieval European villages as well as modern metropolises. And it is alive and well in modern America; average travel time to work falls within a relatively narrow window in cities that look vastly different from each other.
Many Americans assume that dense development leads to gridlock and long commutes. But graphing the population-weighted density of U.S. metro areas against average travel time to work shows at most a weak correlation between density and commute length. Nearly all large metros have travel times between 20 and 30 minutes, whether dense or sprawling, car-dependent or transit-rich.
The densest 20 metro areas have an average commute time of 28.9 minutes. Of the metros that exceed Marchetti’s 30-minute (each way) rule, one is Atlanta, famous for its discontinuous, spread-out development pattern and long freeway commutes. Another is California’s Inland Empire, with a high share of commuters to the adjacent Los Angeles region. All of the others are older coastal metros, including Washington, Boston and New York, that boast relatively high public transit ridership. (Transit commuters may be willing to tolerate longer commutes because they can multitask in ways that drivers cannot.) The metro areas with a less-than-20-minute average commute are invariably smaller ones: not one city in this group exceeds a million residents.
Most of us, if offered an extremely quick means of commuting to work — all else equal — would jump at it. So why don’t transportation-capacity upgrades like new roads and new lanes cause commute times to drop? The likely explanation is that all else is not equal.
People have “travel time budgets” — the amount of time they are willing to spend on the road — and weigh this against other factors when making choices about where to live, where to work, and what time to hit the road if they have a choice. If a road is widened to speed travel, drivers will compensate by rebalancing their calculations. This is likely to lead to increasing travel on new or improved roads until congestion reappears, a phenomenon known in transportation planning as induced demand.
In his 1992 book, Stuck in Traffic, the economist Anthony Downs coined the idea of the principle of triple convergence:
Visualize a major commuting freeway so heavily congested each morning that traffic crawls for at least thirty minutes. If that freeway were magically doubled in capacity overnight, the next day traffic would flow rapidly because the same number of drivers would have twice as much road space.
But very soon word would get around that this road was uncongested. Drivers who had formerly traveled before or after the peak hour to avoid congestion would shift back into that peak period. Drivers who had been using alternative routes would shift onto this now convenient freeway. Some commuters who had been using transit would start driving on this road during peak periods. Within a short time, this triple convergence upon the expanded road during peak hours would make the road as congested as before its expansion.
Not only will short-term driver behavior change, but in the medium to long term, land development patterns will compensate. People will live farther from work or from non-work destinations that are important to them, knowing they have a high-speed connection to those destinations. Homes can be built and sold farther from downtown job clusters. The incentive for outward sprawl, and more and more traffic as a result, persists until travel times restabilize at about the same level as before. In one of the more rigorous academic studies of induced demand, conducted in 2009, Gilles Duranton and Matthew Turner found nearly a perfect 1-to-1 match between new lane-miles of urban freeway and increased travel on those freeways.
The idea of simply paving our way to reduced travel times is thus a non-starter. If average travel times were well above the Marchetti constant, we might have a real structural problem, a situation in which many endure excruciating commutes for the genuine lack of another option. This is the case in some developing-world megacities: the traffic jams of Jakarta and Manila are the stuff of legend.
This kind of structural problem can be solved through either expanding system capacity (allowing faster travel to destinations) or increasing development density (bringing destinations closer to each other), or both. In American cities, though, where the average commute is around 25 minutes each way, the Marchetti constant suggests that drivers would take advantage of the extra capacity to live farther from work or to make more optional trips for shopping and recreation.
Costly expansions of urban freeways would thus at best save commuters a minute or two in the short run. Yet the benefits of such modest savings are often sold to federal funding agencies and the public using mathematical sleight of hand. A few minutes of travel time savings per person, aggregated over many thousands of drivers and multiplied by the prevailing hourly wage, produces impressive dollar amounts of supposedly reclaimed productivity. This is the methodology used, for example, by Inrix in its annual Urban Mobility Scorecard, with time lost to congestion valued at nearly $20 per hour.
But this kind of aggregation makes little sense. If I save two and a half hours in an unbroken chunk, I can use that time in any number of productive ways. Saving five minutes a day for 30 days is not of equivalent value, because I cannot put that 150 minutes, broken as it is into tiny increments, to productive use. Most likely I would not change a thing about my daily routine except maybe snoozing through the alarm in the morning.
This whole-is-greater-than-the-sum-of-the-parts hypothesis is, of course, just that: a hypothesis. But thanks to high occupancy toll lanes that give individuals the option to use the high occupancy lanes for a price, we know that people do not place great value on saving a few minutes here and there. A 2017 study from Louisiana State University estimates that the average user values time savings at about $3 per hour — a far cry from $20.
There is a fair counterargument here. The notion of a constant travel time budget implies that people will defer trips they would otherwise have made to avoid exceeding that budget. If this is the case, then avoided trips will not show up in travel time statistics, but they still represent missed opportunities. More mobility is good, proponents of this argument say, because although new roads will induce people to drive more, that driving is clearly of value to them or they wouldn’t do it.
If this argument were on the mark, we should expect to see congestion bear a cost, not in the form of higher travel times but in the form of an economic drag from the unrealized value of trips not taken. This is a difficult thing to measure, but a simple look at the list of the most congested U.S. metro areas according to Inrix’s index is reason for skepticism. Here’s that list:
- Los Angeles
- New York
- San Francisco
Inrix’s list is virtually a who’s who of places that are prospering, with high employment, high wages and dynamic knowledge economies. That doesn’t mean congestion isn’t taking some sort of economic toll on these cities — it’s possible they’d be even more disproportionately prosperous without it. But to claim the damage is severe would be about as sensible as Yogi Berra’s famous aphorism, “Nobody goes there anymore. It’s too crowded.”
The underlying problem with the “travel is good, so more travel is better” philosophy is that it confuses mobility for accessibility. I can go more distance in less time in Kansas City than I can in New York. I can go even farther in rural Wyoming. But what can I get to in that time? For the sheer amount of opportunity (employment, social, recreational) available in a 30-minute trip, you still can’t beat Manhattan, where Inrix warns us that commuters are limited to a measly 8.23 mph — as if traveling miles were the end and not the means.
We chase mobility like a snake eating its own tail, accruing ever-growing road maintenance obligations so that we can travel more miles at higher speed, live farther afield from each other and ultimately need yet more roads. This arms race with congestion is embedded in public policy and in engineering practice. We measure congestion costs in “delay,” with the implicit assumption that free-flowing traffic — as fast as the road would safely allow us to go — is the ideal. But this logic yields dubious benefits in terms of both economic productivity and personal satisfaction.
What a Traffic Problem Is
There are things to do to reduce the pain of congestion. But we have to think differently, abandoning the idea that our traffic problem is one of delay or travel time itself. We can and should, however, address aspects two and three of our aforementioned three-pronged traffic problem: the variability of commute time and the unpleasantness of the driving experience.
Unpredictability in travel time has significant costs. It is largely caused by extreme congestion leading to breakdowns in traffic flow. In contrast to average time lost commuting, most drivers value reliability highly: toll lane users will pay for reduced travel time variability at the rate of about $23 per hour. It’s important to know you will make it to work on time for the big meeting, or to school in time to pick up your child. Gridlock, which can turn a 20-minute trip into an hour-long trip with little advance warning, is misery-inducing even when average travel times are not.
The most effective tool to combat gridlock on limited-access roads is congestion pricing. Highway lanes can be tolled at peak hours to reduce the use of those lanes. Drivers making unavoidable trips, or those in a hurry, will pay to use them while others will take alternate routes or defer less essential trips until rush hour is over. Toll lanes are most effective in this regard when the fee is not set to meet a revenue goal but to maintain smooth traffic flow, and is adjusted throughout the day as traffic levels change.
Another tool is public transit, which provides an alternative to driving at all — but with a big caveat. I’ve avoided discussing transit thus far for two reasons. First, only a small share of Americans currently live in neighborhoods in which transit is convenient enough to compete with driving in desirability. Second, the evidence suggests transit does little to directly relieve road congestion because induced demand applies: new drivers will fill the road space freed by transit commuters.
Transit does generate considerable benefits in its own right: in compact cities, its capacity and economy of space allow it to move far greater volumes of people than any freeway ever could, and at considerably lower fiscal and environmental cost. But transit may not have a significant role to play right now in congestion reduction in the places where most Americans live.
Even without congestion pricing or major investments in public transit, there is hope in addressing chronic rush-hour congestion in our cities. The key is to think differently about how we handle traffic within the road network. To clarify, let me digress to introduce you to a fellow named Mr. Go.
Mr. Go is the nickname for MRGO, the Mississippi River Gulf Outlet, which was completed by the Army Corps of Engineers in 1965. For decades it provided ship traffic from the port of New Orleans a shortcut through the river’s labyrinthine, muddy delta and into the Gulf of Mexico.
The drawback to MRGO became all too clear when storms dropped colossal amounts of rain on the Gulf Coast region in a short time. The original marshes south of New Orleans were capable of absorbing the resulting huge influx of water, distributing it across hundreds if not thousands of tiny channels, and slowly but steadily filtering it through. These marshes were New Orleans’s best protection against cataclysmic flooding.
The deep, dredged channel of MRGO, on the other hand, would, in flood conditions, become a fire hose aimed squarely at New Orleans. It was a significant factor in the destruction wrought by Hurricane Katrina in 2005, amplifying its storm surge dramatically. In 2009, the channel was closed for good.
Now, a freeway in an urban downtown is like Mr. Go. A complete grid of smaller surface streets, in which there are many equally attractive routes from A to B for drivers to choose from, is more like a wetland, and actually does a great job of dispersing traffic and preventing congestion. Wide, high-speed arterial roads are congestion magnets.
Channeling the vast majority of vehicle trips to freeways rather than to a distributed network of small surface streets does give travelers an efficiency boost at off-peak times. But it almost guarantees that when the freeway “floods” with cars at peak commuting hours, painful gridlock will result.
Dallas-based urban designer Patrick Kennedy is a vocal advocate for the removal of Dallas’s I-345 freeway — a once-longshot project that has gained remarkable political momentum in a city in which the car is still king. His argument for the teardown does not center on a critique of car dependency, but on the notion that the dispersed downtown Dallas grid has more latent capacity to handle traffic than I-345 does, and scrapping the freeway would have the added benefit of reviving a blighted swath of city and reconnecting downtown to adjacent neighborhoods.
There is precedent for such radical action. When San Francisco opted not to rebuild its damaged Embarcadero Freeway after the 1989 Loma Prieta earthquake, the city did not gain a traffic apocalypse but, rather, a revitalized waterfront. When Seoul, South Korea, removed a freeway and restored the Cheonggyecheon Creek through downtown (and created a linear park along its course), property values in the immediate vicinity soared by 30 to 50 percent.
Wide arterials and expressways in an urban setting account for a small fraction of total roadway capacity, but nearly 100 percent of the congestion. These roads enjoy a travel time advantage at off-peak hours, but under rush-hour conditions they offer no advantage over slower, less direct alternate routes.
You can test this — albeit unscientifically — in your own city. Find a location where two routes parallel each other: a high-capacity thoroughfare, and a small, local street with perpetually slow traffic. At weekday rush hour, when the thoroughfare is busy, drive one route, then loop back and drive the other. Compare your times.
In my own (equally unscientific) experiment in Sarasota, I first tried parallel east-west routes at rush hour. One was a principal artery with a 30 mph speed limit and a lot of lights; the other was a two-lane road slowed by angled parking and lots of pedestrian traffic (and a 15 mph speed limit). Travel times proved identical.
I repeated the experiment with an analogous north-south pairing: one four-lane arterial with a 35 mph speed limit and six stoplights, followed by one parallel two-lane urban street with a 25 mph speed limit, three stoplights and a roundabout. In this case, the narrower, “slower” urban street proved to be dramatically faster, cutting my travel time by one-third.
This kind of experiment suggests a solution to congestion that is precisely the opposite of 20th-century engineering dogma: cities would be well served to reconnect their fragmented grids, slow traffic on wide, busy roads, and thus spread out urban traffic instead of concentrating it.
Note, moreover, that this approach not only removes congestion choke points, but that it would be easier to maintain out of local government budgets because it does not involve extensive new construction. It may even raise property values in places where the high-speed, congested thoroughfare has depressed them.
Driving in the marsh city, rather than the Mr. Go city, also has the potential to be more enjoyable and less stressful. There is no comprehensive research on the psychology of driving — which types of street environments make drivers comfortable, and which make them stressed out and jumpy. But there are some rules of thumb we can distill from experiences that should be familiar to anyone who drives a car.
A key one: traveling slowly is not unpleasant. Traveling slowly when you feel you should be able to travel quickly is unpleasant.
Streets have a design speed, which is the maximum speed at which most drivers feel comfortable and safe, based on features such as the width and straightness of the lanes, the presence of visual obstructions such as median trees and signposts, and the amount of open space between the road and adjacent buildings. A mismatch between design speed and attainable speed is psychologically uncomfortable — it’s responsible for the anxiety that accompanies crawling along at 20 mph on a freeway built for 70 mph traffic.
“Traffic calming” is an engineering term for efforts to decrease the design speed of a street by narrowing lanes, adding visual obstacles and generally creating an environment that induces drivers to travel more slowly without the need for stepped-up speed enforcement. As the Sarasota experiment shows, two traffic-calmed streets at rush hour were able to transport me and my car at the same effective speed — about 12 mph — as two much larger thoroughfares with much higher design speeds. And the experience was safer for pedestrians and cyclists I encountered, and more pleasant for me.
Roundabouts are another low-cost intervention that would help smooth traffic flow in many urban places. They facilitate the traffic-calming effect by allowing travel at a steady 15 mph instead of the stop-and-go of accelerating to 30 only to brake at the next red light or congested stretch. They are less expensive than stoplights to maintain and operate, and the Federal Highway Administration estimates they can reduce crash rates by 70 percent or more.
These changes are low-cost and incremental. But the end result of enough of them would be a very different type of city — one characterized by high connectivity, with many routes between any two points. This different kind of city would lack the clear hierarchical distinctions between neighborhood streets and through streets that exist in many places today. That hierarchy — a regional road network that channels most traffic onto a handful of large, high-speed roads like flood waters into MRGO — is the fundamental cause of much urban traffic congestion.
The marsh city, rather than the MRGO city, is not very conductive to long-distance commuting. But a more compact city and suburbs with a well-connected grid offer high accessibility, not high mobility. You can get anywhere, from anywhere, by a large number of different routes — albeit not at high speed. Marchetti’s constant suggests that changes of this ilk would be accompanied by changes in land use over time: an increased demand for denser cities and less development on the far suburban fringe as residents seek to keep their overall travel time “under budget.”
Charles Marohn of Strong Towns, an organization that advocates an incremental, fiscally sustainable model of urban development, observes that the most effective long-term congestion remedy involves these changes in land use:
For nearly seven decades, our national transportation obsession has been about maximizing the amount that you can drive. We now need to focus on minimizing the amount you are forced to drive.
No city can be expected to transform its land-use patterns overnight, and opportunities to alter the street grid are rare because of vested property rights. However, cities can, at minimal cost, require that new development feature a highly connected street grid. And they can seek out opportunities to create such connections where they do not currently exist — for example, when large urban sites are redeveloped. Finally, they can use the techniques of traffic calming to distribute the burden of rush-hour traffic across a greater number of routes.
The result for drivers would not be less time spent in traffic — Marchetti’s constant suggests that is an unattainable goal. It would, however, with any luck, be less time spent miserable in traffic.