Traffic Congestion: Why It’s Increasing and How To Reduce It

The statistics show that each of us is driving less.  So why do our roads feel more jammed up?  Why does it take longer to get anywhere?  And what can we do about it?  Some politicians have begun blaming Traffic Calming and bicycle lanes for the backups; saying that Complete Streets and pedestrian bulb-outs are making roads less safe because less accessible for emergency vehicles.  Is there any truth to this?  More fundamentally, is car congestion a problem to be solved or a solution to a problem?

A 2013 report from US PIRG showed that the average number of miles driven by the average American has been falling for about a decade, through economic booms and busts, and was down to mid-1990s levels.  Millennials, our nation’s largest-ever generational cohort, are using transit and bikes more and taking fewer and shorter car trips, resulting in a 23% drop in the average number of miles driven.  The percentage of high school seniors with a driver’s license fell 12%.  Walkable city life is increasingly attractive to both young people and retiring baby boomers.  The rise of on-line shopping, social media, and telecommuting has meant fewer quick car trips.

Despite these trends, as every driver knows, our roads are increasingly congested – not everywhere or all the time but for increasing periods at a growing number of key intersections and road segments.  Congestion radically reduces the volume of traffic passing through a road section, the through-put, thereby creating a negative feedback loop that creates more backups.   It’s estimated that USA drivers spend about 14.5 million hours every day stuck in traffic.  Congestion not only costs us time – in 2011 Boston drivers collectively lost about 137 million hours, or about 53 hours per commuter per year – but also fuel and therefore pollution, health, and money.  Not to mention frustration and occasionally murderous road rage.  Although we Bostonians believe we’ve got it worst, car congestion seems to be clogging roads like kudzu in nearly every city in the country – and, by some reports, across the globe.

It’s true that a new report has said that the first four months of 2015 has set a new record in total vehicle miles in the US – up nearly 32 billion since the previous high in 2007, pushing gas consumption as well as prices upward. Lower gasoline prices and a recovering economy (consumer spending in May, 2015 had the highest month jump in six years) are two reasons for the jump, probably augmented by the continuing lack of viable alternatives to car driving for many people.  But a four-month blip is not enough to explain years of delays.

We do know some things that are contributing to the larger problem – land use patterns and population growth are the most important.  The low-rise dense designs that make older urban areas walkable and transit-efficient is illegal to build in many places today due to parking requirements, anti-mixed use and other zoning requirements, etc.

We know some things that may appear to be causative but actually aren’t – making roads safer for pedestrians and bicyclists, prioritizing bus and trolley traffic, even reducing the average speed of cars.

We know some things that (counterintuitively) do not help reduce congestion – most notably building more roads or adding lanes, all of which eventually fill up as our additional drivers decide to move into the new space.

And we know some things that do improve the situation, but usually only when they are applied as a group rather than singularly – improving road use efficiency using technology (signal timing, access controls, central monitoring) and other methods (car pools, HOV lanes, car sharing, perhaps driverless cars), increasing alternative options (transit both regional anddowntown, bicycling), changing land-use patterns (Smart-Growth style transit-orientated development), requiring corporate and municipal Transportation Demand Management programs (incentives to not drive alone or to not drive at all), and (most effective of all) congestion pricing of various kinds.

What is needed is the cultural and political willingness to accept this knowledge and act upon it – while also coming to grips with the reality that the continuing imbalance of potential drivers to current or any plausible future amounts of road space means that congestion is a permanent part of a car-based reality.



We tend to think of congestion as something that happens on a stretch of road, occurring when the number of cars is higher than the road’s capacity, forcing everyone into stop-and-go movement.  But on-road jams are even more likely to be back-ups from pinch-points – intersections with poor signal timing, lane merging, accidents, rubber-necking when going past something interesting, and other constrictions – that suddenly create rolling waves of braking and over-capacity density along an otherwise functional road.  One analysis says that up to 40% of congestion is caused by “bottlenecks,” about 25% by “traffic incidents,” and 10% by “work zones.”

The US population has increased by about 33% over the past fifty years.  However, as our national wealth has increased the number of registered cars grew nearly three times as much, 90%, during the same period.  Even though we are each driving less, there are so many more of us that the total annual vehicle miles traveled is about 130% higher than in 1970.  On the other hand, despite a recent 50% surge in new highway construction (the resulting disruption is another reason for today’s perception of increased backups) the total number of road miles has grown by only 6% during the past half century – and may even shrink in the future.

Not only are there more of us, we are more spread out; instead of building up, we built out.  Its dumb growth, or at least wasteful.  From 1960 to 2010 metropolitan areas in the USAtripled in size to 912,992 square miles, approximately a quarter of the total land area.   The process has pushed development farther into the hinterlands leaving decay behind it, like a sprawling forest fire that has consumed a quarter of the country’s land area. The trend has continued, population-weighted population density fell by 16 people per square mile between 2000 and 2010, while in metropolitan areas it fell by an enormous 405 people per square mile despite the urbanization trend (one theory is that incoming city dwellers are more affluent and demand more space per person).  Many people do not want to drive and do not even want to live where they need a car, but have no alternative because the only housing they can afford is in an auto-oriented suburb.

However, our employment and shopping centers are more concentrated.  So our our road system is inherently designed like a watershed – lots of small brooks collecting into bigger streams and then into raging rivers.  As our population expands into once rural regions the car traffic pressure on collector roads and bridges leading to large employment, shopping, and recreation centers inevitably increases.  Ironically, the recent trend towards urbanization doesn’t automatically help the situation: to the extent that the new city dwellers are still driving, their presence puts even more concentrated pressure on key intersections and other choke points.


The core problem is that there are simply many more people potentially able to drive than our roads have the capacity to handle given our current land use patterns based around a hierarchical road system.  Congestion can be seen as a way to ration our limited road space – a solution to the problem of over demand.  In market terms, the solution would be to increase the supply:  in transportation terms that means build more roads.  But, as with so many market-based solutions, this ignores the side effects, the externalities.

Once upon a time, increasing road capacity, and therefore traffic volume, was believed to lead to so much economic growth that potential side effects were irrelevant.   However, we’ve now learned that the externalities of highway construction are enormous – devastated neighborhoods, economic decline, and escalating health problems in urban areas; water problems, decreasing regional food self-sufficiency, and sprawl in rural areas.   The side effects of increasing road capacity are so bad that several states, most notably environmentally-challenged California, have begun exploring no longer using the impact on car Level of Service (LOS) as a criteria for evaluating new developments or establishing mitigation requirements. (The emerging alternative is to require developers to pay for ways to handle the increased number of trips the project will create, but allow lower-cost non-car modes to be counted towards the total.)

But even if we had the space and the money to create the unimaginable amount of new pavement needed to clear the roads, it would be a short-term victory.  Like a gas expanding to fill all available space until the pressure equalizes across the system, drivers will change their plans to take advantage of any quicker and easier routes created by the new capacity.   If the cost to the driver in time and frustration of current congestion is reduced more people will choose to drive.   In a process called “triple convergence” any improvement in driving conditions on a particular route will attract drivers “from (a) other times, (b) other routes, or (c) other modes of travel.”  The third factor explains why building new roads also creates more drivers – which then hastens the filling up of the new facility.  And fixing one part of a congested system simply moves the problem to the next segment, as the traffic pileups on either side of the smooth-flowing Big Dig tunnels prove.

(Good news:  the “if you build it they will come” dynamic also works in reverse – “if you take it away they go away” through a “triple deconverging” process back to other times, routes, or modes making cars disappear from the transportation system to a degree that current traffic prediction models consistently underestimate.)


One solution is to make better use of the roads we have.  High Tech firms are salivating at the opportunity to sell “Intelligent Transport Systems” (ITS) as part of even bigger “smart city” systems.   Embedded sensors.  Video monitoring cameras.  Adjustable traffic signals that change the timing or divert people into alternative routes.  These are extraordinarily expensive, require enormously skilled operational refinements and maintenance after installation, and lock cities up into endless (and also expensive) upgrading.  At the price of personal privacy, new apps take car location and speed and let drivers detour.  Driverless cars may someday allow cars to be more closely packed on roads without causing accidents.

Reducing the number of cars on the road is a complementary strategy.  The spread of car-sharing services, both for-profit and publicly owned, has gone along with the reduction in car ownership.  Some studies indicate that each shared car takes between 10 and 25 cars off the road.  Public transit is even more powerful.  The Texas A&M Transportation Institute has found that without public transit, USA car drivers would have an additional 785 million hours of delay each year.  In Boston, the MBTA saved regional drivers nearly 32.9 million hours of road stall in 2010.  The number of cars one full bus eliminates from the road is so large that increasing the amount, effectiveness, and attraction of public transit is a powerful strategy even if it requires creating reserved bus (and HOV) lanes during rush hours.

Counterintuitively, in addition to reducing our still-appalling level of road kill, a similar traffic improvement effect comes from Traffic Calming (using a variety of techniques to slow cars to a desired speed) and Complete Streets (giving equal priority to facilities for transit vehicles, bicyclists, and pedestrians as well as cars and trucks).  A pedestrian is killed every 2 hours in this country, a moral outrage that the growing number of Vision Zero campaigns are beginning to take seriously.  These safety-enhancing strategies are often blamed for congestion because they visible reduce road widths, reduce speeds, and sometimes eliminate lanes.  But study after study shows that these first impressions are not only wrong, but backward, at least in urban areas.  In New York City, for example, the creation of protected bike lanes has also created space for left-turn pockets, which means turning cars no longer block the through lanes and traffic flow has actually improved despite the loss of a lane.

Even when creating left-turn lanes isn’t possible, improving safety generally helps car flow.  First, by making non-car modes safer and more attractive, these strategies reduce car traffic and free up parking – partly by facilitating mode shifts among current residents but probably also by tilting the mix of new residents towards people looking for a place where they can walk or bike rather than drive.  (Reducing VMT is a particularly effective strategy for reducing car-related injuries – the fewer miles people drive the fewer people get hurt.  This is the opposite of bicycle riding, where increasing number lowers the rate of injury and whose public health benefit increases along with increased participation.)

Second, high speed driving in cities is usually the result of driver impatience and seldom lowers the time it takes to get through any trip of more than a few blocks.  Urban areas simply have too many obstructions, the high speed creates danger where it happens but it can’t last long, it’s simply “racing to the next red light.”  In Somerville, the top speed on Beacon Street was 30 mph but during rush hour the average speed was a little above 10 mph.  This speed had little connection to the width of the road or of a crosswalk.  Using road space for improved pedestrian and bicycle safety has little impact.

Third, the key to traffic efficiency is throughput, the total number of cars that can get through a particular distance in a set amount of time, and just like Aesop told us “slow and steady wins the race.”  The trick is sophisticated traffic-signal timing that varies according to the traffic load on multiple streets at different times of day – something that ITS might facilitate but that few cities currently have.  These systems also malfunction without constant maintenance of both software and fixtures, a level of skill and expense often beyond the capacities of municipal DPW’s.  And a malfunctioning signal can lead to more congestion than having no signal at all.   But reducing theoretical road capacity is not the issue:  Cambridge, for example, cut three lanes out of Mass Ave in Central Square to install bike lanes and while traffic is definitely slower the total number of vehicles passing through per hour has not been reduced.  The current installation of a protected bike lane on Western Ave. in Cambridge will similarly have negligible effect on throughput compared with signal timing (although the construction process has driven everyone crazy).


Making it more expensive to drive will also reduce traffic, but it is not as straightforward as first appears.  European countries have hugely higher levels of sales tax on car purchases, annual excise tax for ownership (as a surrogate for usage), and per-gallon gasoline taxes – all of which does provide a market-based incentive to not own or use a car to the extent possible.  A complementary approach is to limit the number and raise the cost of parking spots in congested areas.  But unless these are offset by a parallel but not directly connected annual tax credit, these costs fall heaviest on those who have no choice other than using a car – and on the low-income subset in particular.  Still, raising the cost of cars and driving while reducing its convenience is a powerful “nudge.”  Unfortunately, there is little public appetite for this self-inflicted pain as shown by the recent overwhelming passage of a Massachusetts ballot question canceling the proposed increase of gas taxes to keep up with inflation.

Another car-use fee is tolls.  There is a lot of discussion these days about letting private firms build new roads and recoup their costs through tolls, such as the proposed third Cape Cod bridge or the dangerous conversion of the Route 3 breakdown lane to a toll road.  If the cost of the tolls is high enough, these plans would create an open road for those able and willing to pay.  However, it would have absolutely no positive effect on everyone else.  To the extent that the new roads reduce traffic on the regular roads, the “triple convergence” effect would simply attract additional drivers to fill the new space; the non-toll roads would be just as jammed as before.  The net effect would be to use valuable land for the benefit of the rich, perhaps enticing even more of them to drive and thereby reducing their influencial  support for mass transit.  (An unintended consequence of this approach is to increase the “export” of money out of the local economy:  traffic on Virginia’s Routes 495 and 95 move faster than the adjoining “free” lanes, and are free for carpools, but the money goes to an Australian company.)

The only way to avoid this is to limit privatization to currently existing roads – requiring private firms to “buy out” public ownership and invest as much as needed to upgrade the road quality in exchange for the right to charge a regulated level of tolls on all or some lanes.  This would also hurt those with the fewest choices and least able to pay, but because it wouldn’t create any new road capacity it wouldn’t attract any more drivers on to the roads – in fact, it might push some percentage of current drivers to find other routes or other methods of travel (or to change where they live or where they work).  However, even better than turning lanes into toll roads would be turning them into express bus (and carpool) priority lanes.  AsEnrique Penalosa says, “Congestion is the signal that you need to expand public transit.”  A high quality bus rapid transit system would provide a non-regressive alternative to using a car.


The most fundamental solution to congestion is changing land use patterns, both in the city and in the suburbs.  The young professionals now adding to the gentrification of many neighborhoods will eventually have families and want more space (as well as better schools).  The only choice they’re currently given is sprawling auto-centric homes far from the city.  Design of suburban neighborhoods does not need to be car-centric, they can be walkable and still provide space and privacy, as seen in many early 1900’s streetcar suburbs that are designed to be walkable to everyday needs and transit.  What about duplexes, triplexes, row-houses with a common interior-block yard, mother-in-law units on single-family lots?  Many models other than single-family home offer open space, a significant level of privacy, and also build community, increase safety, and make transit and walking much more viable in new developments and for a changing demographic in post-war neighborhoods.


All these strategies help reduce congestion, but only a bit and only if they are done together.  The reason for their weakness is the “triple convergence” problem:  to the extent they reduce car traffic the huge pool of potential drivers looking for the most convenient, quickest, and cost-effective method of travel will divert themselves into the suddenly attractive road.

However, these strategies are vitally important because they lay the foundation for the strategies that do have deeper impact:  changing drivers’ decision-making context throughTransportation Demand Management (TDM) programs and raising the general cost of driving through tax policies or Congestion Pricing.   TDM is typically focused on destination owners – employers and shopping centers – and tends to be somewhat more progressive;  Congestion pricing tends to be user focused and therefore regressive.   The key point, however, is that the impact on the driver of these market-based approaches stay in effect even as road traffic lowers, continuing to tilt traveler’s decision-making away from driving alone.

The Cambridge TDM program emerged from an EPA air quality suit against the city and focuses on new commercial development.  Any project that adds more than 5 parking spots has to also set up a program to actively encourage non-car commuting (subsidized transit passes, bike parking, car pools, etc.) and forbids providing the typical driving inducements (free or even any parking, etc.)  Mostly as a result while the booming Kendal Square High-Tech and Bio-Tech area has seen a 40% increase in commercial and institutional space over the past decade, vehicle counts have fallen by as much as 14% on major streets.

Also proven effective is congestion pricing – charging a fee to use certain roads or enter certain areas depending on the level of demand:  the more people who want to use a road at a particular time the higher the toll.  It shifts traffic by 10% to 30%.  Because upper income people tend to drive more and be less sensitive to tolls, it has been claimed that the imposition of congestion pricing counterintuitively reduces the relative tax burden on lower income families. (Providing additional tax credits can also relieve the burden while maintaining the congestion-reducing impact of the use-tax.)


Car owners have been expressing their anger at having to slow down for others and at congestion almost since cars first appeared in our streets about a century ago.  We don’t like slowing down. We don’t like waiting.  I remember my parents complaining about traffic in the 1950s.

But so long as our population increases and we are wealthy enough for large numbers of people to afford cars, and so long as our land use continues to force a dispersed population to assemble themselves in a relatively few locations to work and shop, the capacity of our roads will not be able to expand enough to prevent increasing levels of congestion at pinch-points of various kinds.

The best we can do is create attractive alternatives to single-occupancy travel and to the use of cars for short trips.  Mass transit is the foundation. Car-share, bicycling, walking, and the like are the necessary finishing touches.   And then we have to have the political and personal will to raise the cost of driving and parking – in time, convenience, flexibility, and money – relative to these alternatives.   The good news is that reducing congestion in this way helps raise the money needed for the alternatives but also has a long list of equally valuable positive side effects from reduced pollution and greenhouse gas emissions to losing less money to non-local fuel suppliers, to increasing safety, to making our communities and region more livable.

It might be worth the effort.


Thanks to Jason DeGray, Alex Epstein, and Jamie Young for comments on earlier drafts.  All remaining errors and opinions are, of course, my responsibility.




Showing 2 reactions

Please check your e-mail for a link to activate your account.
  • commented 2015-12-07 22:14:38 -0500
    The key point of this blog was that there is no car-based solution to car-traffic congestion – even if we each continue to drive less there are many more of us with many more of us owning a car, so total annual VMT continues to climb and we have more potential drivers than road capacity. Given our residential sprawl around our relatively concentrated employment/commercial centers it is inevitable that congestion arises at every pinch point and every increase in road capacity merely attracts more drivers to fill it. We should stop using congestion reduction as a reason for road work – the only worthwhile investments are for safety and to increase options for non-single-occupancy-car-travel.

    As it turns out, the day after this blog was emailed out I attended a conference at the BU Initiative on Cities about how to “nudge” people towards better (safer, more efficient, less costly) transportation behaviors. One speaker reminded us that 25-30% of downtown driving is people circling the block looking for parking. So parking reform – specifically demand-based pricing perhaps combined with software directing people to available spots – might reduce that type of congestion. Another speaker said that up to 25% of his city’s highway congestion was backups caused by car accidents – so a regional Vision Zero campaign eliminate traffic fatalities and serious injuries might help move cars along major roads. A third speaker pointed out that every additional car-sharing vehicle (Zipcar, Lyft, and others) removes 12 to 25 private cars from our streets. Most interesting were the speakers who described programs to improve decision-making and driving behavior using education, skill-training, gamification, social norm setting and status adjustments, peer support, and material rewards.. It seems to work, at least in the short run. But as Public Health professionals learned through the fight against smoking, it takes a careful combination of education and positive supports along with making bad behaviors more difficult, expensive, and less social accepted through regulation, media/cultural messaging, and market mechanisms. (See my previous blog: What Transportation And Public Health Can Learn From Each Other About Changing Public Behaviors )

    It wasn’t clear if a perfect combination of all these strategies would outweigh the congestion-causing factors. (There were several suggestions that the much-headlined future arrival of driverless cars will actually make congestion worse!) But a perfect combination would be hard to create. I stand by my point that we can only make car driving better by focusing on improving our non-single-occupancy-car-travel options.

  • commented 2015-12-06 15:12:48 -0500
    Thanks for this Steve! I’m ready to join any serious push for congestion pricing and bus rapid transit in Cambridge. It’s high time!