Traffic Tipping Points
This comes from CEOs for Cities:
Last year, the US made more progress in reducing traffic congestion than any other time in history. New data show that in 2008 the amount of traffic congestion in the nation’s cities declined by 30 percent, and that congestion was lower in every hour of every day in 2008 than it had been the year previously. How did we make these big gains? Not by adding more highway lanes or transit — the physical infrastructure barely changed — we did it with a very modest decline in car travel. On urban interstate highways, total vehicle miles traveled in the US declined by about 3 percent in 2008…
The decline in congestion — which analysts have labeled ’startling” — was almost universal. Traffic congestion actually declined in 99 of the nation’s 100 largest metro areas, according to Inrix, which monitors traffic around the nation…
How did such a small decline in travel produce such a big drop in congestion? It’s well known that traffic congestion is subject to a tipping point–what economists call non-linearities. Add an additional car to a crowded road at rush hour, and traffic slows down a bit, and then the “carrying capacity” of the road declines. Traffic engineers estimate that most roads carry their maximum throughput — number of vehicles per hour at about 40 miles per hour — so as traffic slows below that speed, the road actually loses capacity and goes slower and slower, producing a traffic jam.
In other words, roads work really well up to a point, and then they work really badly. And many, many roads in America work really badly much of the time, costing Americans some $78 billion in lost time and wasted fuel. But high fuel prices led to reduced driving last year — enough reduced driving that many roads which had been working badly began to work well again. But because of the tipping point phenomenon, it didn’t take anywhere near a 30% reduction in driving to get a 30% reduction in congestion.
This is the intuition behind congestion pricing — that there are big losses to adding cars to a road beyond a certain point, and by limiting the number of vehicles on a road through a variable toll, capacity is maximized. And obviously, this process is made more efficient (as is the process of substitution away from high fuel prices) if quality substitutes are available — like public transit, or bike lanes, or affordable housing within walking distance of employment centers.
What people so often miss is that these policy changes make roads more effective and driving more efficient; they’re good for drivers. That they also happen to be good for the environment and for the economy is just a nice side benefit, as far as drivers need be concerned.
March 3rd, 2009 at 6:17 pm
What this also means, of course, is that you can potentially achieve a big reduction in road congestion with only a small increase in road capacity.
The dramatic decline in congestion between 2007 and 2008 also undermines the case for more transit on congestion-relief grounds.
March 3rd, 2009 at 7:08 pm
Question for Mixner: How do you get a
“small” increase in road capacity? Add one more lane to an existing freeway? Investment in highway capacity is lumpy (economists say indivisible); you can’t expand a highway by 10% of a lane.
Small increases in capacity are now hugely expensive in congested urban areas. One five-mile freeway widening/bridge project is estimated to cost in excess of $3 billion in my town.
Americans line up in traffic for the same reason Russians (used) to line up for bread: the price was set too low. Fix the pricing and the lines will go away and the quality will improve.
March 3rd, 2009 at 7:36 pm
Question for Mixner: How do you get a “small” increase in road capacity? Add one more lane to an existing freeway?
If the road in question is a freeway, yes, that’s one way. Or you could build or expand other roads to divert some of the freeway traffic.
Small increases in capacity are now hugely expensive in congested urban areas. One five-mile freeway widening/bridge project is estimated to cost in excess of $3 billion in my town.
Urban rail transit is also hugely expensive. The point is that if you only need a small increase in road capacity to achieve a large decrease in congestion, expanding roads may be a far more cost-effective way of reducing congestion than expanding transit.
Americans line up in traffic for the same reason Russians (used) to line up for bread: the price was set too low.
No, the reason Russians lined up for bread (and most other consumer goods) is because the supply was too low. Congestion pricing of roads may make sense in some cases, but in general congestion is a symptom of inadequate supply.
March 3rd, 2009 at 7:40 pm
This “tipping point” explanation of congestion pricing is very good. We have been doing outreach and education on congestion pricing and have found it is a difficult concept for non-economists to grasp. The public has become so accustomed to what they perceive as “free roads” that it is hard to convince them that pricing is the only effective way of managing congestion and getting the most out of expensive urban highway systems during the few areas a day when they are congested.
March 3rd, 2009 at 8:03 pm
Mixner:
Show me the “small,” cheap improvements in road capacity.
Brush up on your Russian economic history. The state regulated the price of bread so low that it was cheap enough to use it for animal feed. It also had to heavily subsidize farmers, who had no incentives to increase supply. Supply was too low, and demand was too high because the prices were wrong.
March 3rd, 2009 at 8:48 pm
Show me the “small,” cheap improvements in road capacity.
It could be something as cheap as a single-lane local road that bypasses the congested business district of a small town or as expensive as a new freeway in a large urban area. “Small” doesn’t necessarily mean small in terms of its absolute dollar cost. It means small in relation to the benefit it would produce in congestion relief. Because of the tipping-point phenomenon, a small increase in the ratio of supply to demand could produce a large benefit from reduced congestion.
Brush up on your Russian economic history. The state regulated the price of bread so low that it was cheap enough to use it for animal feed. It also had to heavily subsidize farmers, who had no incentives to increase supply. Supply was too low, and demand was too high because the prices were wrong.
Whatever the case with bread specifically, my point is that consumer goods in general were in short supply in Russia. The government didn’t produce enough of them. In the United States, distortions of supply and demand created by subsidies is a much bigger problem for public transportation than for roads. On average, transit users pay only about 27% of the cost of providing the service they use. That encourages massive overconsumption. What’s worse, pricing is usually independent of the distance of travel as well as congestion.
March 4th, 2009 at 8:47 am
Mixner says:
What this also means, of course, is that you can potentially achieve a big reduction in road congestion with only a small increase in road capacity.
Mixner, a little bit of knowledge is a dangerous thing. Yes, adding a bit of road capacity would theoretically reduce road congestion, but only for about one day. The next day, as soon as people knew about the reduced congestion, (1) people would flock to the uncongested road from other roads, known as “spatial convergence” (2) people would start driving on the newly-uncongested road at rush hour, who would otherwise have avoided driving at that time, known as “temporal convergence” and (3) people would switch from transit back to driving, known as “modal convergence”. Check out the book “Still Stuck in Traffic”, to learn more about what the author calls “triple convergence”.
In other words, improving roads never, ever, reduces congestion in the long run. If you improve roads, more people will drive on them, and more importantly in the long run, more development rises along the roads.
The only ways out of congestion in the short run are toll roads that keep a few people off of the road at rush hour, and in the long run, to (1) tax gasoline by several dollars per gallon as Europe has done for years to pay for the military expenses, 42000 traffic deaths per year and pollution and sprawl costs (2) make sure people in suburbs pay much higher taxes and costs for sewer, water, electricity, fire protection, etc that reflects the enormously higher costs involved in providing those services to people that live in those ridiculously-low-density areas.
March 4th, 2009 at 1:16 pm
Russell,
Congestion declines when capacity increases relative to demand. This can happen through either a decrease in demand or an increase in capacity. Your claim that we cannot relieve congestion by increasing capacity is just factually incorrect. Demand is constrained by population and density as well as cost.
Ed Glaeser has found that using a midrange estimate of the unit cost of carbon, the average difference in the cost of carbon emissions between suburban households and central city households is on the order of $200 per year. This is only a small fraction of the differences in the cost of living between suburbs and central cities. Housing cost differences alone far exceed differences in carbon cost. So a carbon tax would be unlikely to have a significant effect on the housing market. If you think you have evidence that central city households subsidize the provision of public services to suburban households (sewer, water, electricity, fire protection) please present it.
March 4th, 2009 at 2:04 pm
Mixner, demand is not constrained by density or by population. VMT has been increasing at a far faster rate than population growth. Looking at the last 25 years, people are driving more, they’re taking more trips, they’re taking longer trips, and they’re using cars for more trips. Population increase only accounts for a very small portion of the total increases in VMT - something on the order of 15-20%.
The simple fact is that adding road capacity does very little to address those capacity concerns in the short term. In the long term, adding road capacity does a lot to increase the other factors - low density land use, autodependent development patterns, etc. These factors are self-reinforcing. You can never catch up. You can never build your way out of congestion, because each additional unit of capacity induces more traffic than it can handle.
March 4th, 2009 at 2:33 pm
The fact that demand has increased faster than population growth does not mean that demand is not constrained by population. It simply means that demand per capita has increased. If population growth had been higher, demand would most likely have been higher also. And of course congestion is not simply a matter of aggregate capacity and aggregate demand, but the distribution of demand across capacity. The higher the geographical concentration of demand, the more congestion there is likely to be. That is why reducing density also tends to reduce congestion. When traffic is distributed more evenly across the road system, congestion will go down. Again, this has been studied academically. As Ed Glaeser reports in his study Sprawl and Urban Growth:
March 4th, 2009 at 5:40 pm
Mix, you apparently either didn’t read Glaeser’s conclusion or are trying to mislead: “Either we need to be accustomed to traffic jams, as Tony Downs (1992) eloquently argued in Stuck in Traffic, or we need to figure out how to be smarter about using electronic tolls to limit congestion.”
March 4th, 2009 at 8:49 pm
I didn’t say or suggest that Glaeser opposes all use of congestion pricing. I was rebutting the claim that congestion cannot be addressed by increasing road capacity or reducing density.
March 5th, 2009 at 3:03 am
Russell,
The only ways out of congestion in the short run are toll roads that keep a few people off of the road at rush hour, and in the long run, to (1) tax gasoline by several dollars per gallon as Europe has done for years to pay for the military expenses, 42000 traffic deaths per year and pollution and sprawl costs (2) make sure people in suburbs pay much higher taxes and costs for sewer, water, electricity, fire protection, etc that reflects the enormously higher costs involved in providing those services to people that live in those ridiculously-low-density areas.
I don’t think this is right. The only way to deal with congestion in either the short or long run is to price roads properly.
Gas taxes don’t do the trick because the marginal cost of an extra mile is the same regardless of the amount of traffic. High gas taxes don’t make drivers take into account the cost they inflict on others. (M4 west of London is notoriously congested despite the high gas taxes there. So is Rome.)
Charging suburbanites higher prices for sewer, water, etc. won’t work either. Raising the price of sewer hook ups won’t give suburbanites an incentive not to drive on congested roads.
Higher prices for these things certainly won’t cause suburbanites to decamp to the central cities en masse. They won’t even necessarily lead to many fewer subdivisions. The reason is that undercharging for sewers, water, etc. is really just a subsidy for the owners of undeveloped land. Eliminating the subsidy might cause some property owners to keep the land as farmland, but you’re talking about a reduction at the margin; some land will still be more valuable as residential than farmland.
My point is that pricing these things properly won’t eliminate suburbanites or suburbanite drivers. We still have to price the roads properly.
March 9th, 2009 at 2:02 am
Mixner: Has trouble acknowledging that turning a crowded 4 lane highway into a free-flowing 6 lane highway almost always results, in the long run, in a crowded 6 lane highway. In cheap-gas America.
Basically because of Mix’s theoretical thought that since there are limits to population and population density in a region, surely with enough lanes on any given road you can get ahead of that finite pool of drivers once and for all.
Hey, apparently my post was too boring to actually read, for which I apologize. Upon actual reading Mixner could have seen the THREE PLACES where I point out all the new drivers come from, NONE of which are from a growing populace or population density. The new drivers that rush onto newly widened arterial roads come from (1) other roads (2) other time slots of travel and (3) other modes of transit, NOT FROM A RIDICULOUSLY EVER-GROWING POPULATION.
Thus you could have all the first-born of the empire sent off to foreign wars, and push all the elders onto ice floes, and limit the distribution of salt thoughout the land, but a 4 lane congested highway that gets widened to 6 lanes will still recongest in short order. There’s just forever too many less convenient roads, or ways of travelling, that surround the newly widened artery, that make it too delicious to resist. A little like the way water flows happily through a hole punched at the bottom of a bucket, even if there are a lot of really big holes already punched at the top.
Mixner further:
If you think you have evidence that central city households subsidize the provision of public services to suburban households (sewer, water, electricity, fire protection) please present it.
Well in plain old Cincinnati Ohio the folks living on 4 acre lawn lots 8 miles north of downtown pay the same per-volume monthly sewer rates as the people in 8 unit apartment buildings, apartments located 2000 feet from an urban core pumping station serving zillions of people. Neither do the downtowners get any tax breaks because many of the police are on bicycles rather than those living-room-sofa-sized police cruisers.
Neither do the central core dwellers get any county environmental air quality credits off their tax bills for the on-average much lower amount of hazardous gasoline, particulates and CO2 they use or produce yearly. (By way of documenting, let’s note that suburbanized Atlantans use over 3,000 gallons of gasoline a year in various ways, and folks in super dense Barcelona use less than 100).
The untaxed costs to a city, due to arterial freeways to suburbs, in terms of ugliness, noise, shattered neighborhoods and residences far, far removed from workplaces, are to me and sociologist Jane Jacobs self-evident, but because unquantifiable might not constitute “evidence” of costs to you.
AC says: I don’t think this is right. The only way to deal with congestion in either the short or long run is to price roads properly.
AC, there’s no disagreement, just an informal language difference here. When I said the only way to get people off of roads at rush hour is “toll roads”, I used a perhaps-now-oldish-phrase to mean the general concept of charging at least some people in some lanes enough money to thin out traffic on a given stretch of road at rush hour–what you are referring to as road pricing. We agree.
March 9th, 2009 at 12:36 pm
Russell,
Do you have any evidence to support your (highly implausible) assertions regarding induced demand and utilities pricing?
March 9th, 2009 at 10:13 pm
Mixner,
Not sure what you mean by “assertion regarding induced demand and utilities pricing.” Please give a bit more detailed or perhaps a tiny specific example of your request.
March 9th, 2009 at 10:53 pm
For starters, this assertion:
“turning a crowded 4 lane highway into a free-flowing 6 lane highway almost always results, in the long run, in a crowded 6 lane highway.”
and this one:
“The new drivers that rush onto newly widened arterial roads come from (1) other roads (2) other time slots of travel and (3) other modes of transit”
and this one:
“in plain old Cincinnati Ohio the folks living on 4 acre lawn lots 8 miles north of downtown pay the same per-volume monthly sewer rates as the people in 8 unit apartment buildings, apartments located 2000 feet from an urban core pumping station serving zillions of people”
Of course, even if you substantiate those assertions, it wouldn’t support your broader claims, anyway.
March 9th, 2009 at 11:05 pm
AC,
The only way to deal with congestion in either the short or long run is to price roads properly.
Then how do you explain the huge differences in congestion between different metropolitan areas? MSAs with little congestion aren’t achieving that through road pricing. They’re doing it through lower population densities and more decentralized use of land.
March 11th, 2009 at 11:20 pm
Mixner,
Why are you bringing up all the uncongested roads in the world, as if I was asserting that “all unpriced roads are congested”? The topic is (or at least I have been thinking it is) how to alleviate congestion on highways that are already congested. I don’t have to explain why there are uncongested roads in the world, neither you nor I have been talking about such trouble-free roads. The topic is, can you widen a congested stretch of highway and thereby eliminate congestion for a long time. You say yes, because there’d be plenty of room for all the traffic. I say, no, almost never, because soon enough new traffic always rushes onto such newly widened roads from other roads, other transit modes and other transit times. As documented well in the book Still Stuck In Traffic.
MSA’s aren’t “achieving” low congestion of already congested roads through lower population densities. Your statement is misleading. Yes, their roads might be forever uncongested thanks to low population densities, but our topic of discussion is fixing a congested road, which I claim you can’t do by widening the road.
Then I guess you’re saying, “prove it, Russell”. Hmm, I would ask you to examine your own assertion, which is when (a) a place is so attractive that it has attracted too many occupants, you can (b) then make it even more attractive, and (c) no new users will be drawn to the place.
I would say that your premise that making a highway more attractive won’t attract new drivers from other roads and times and modes is so inherently at odds with ordinary logic and experience, that I don’t think I need to list every widened highway in the U.S. that then got recongested to make the point. I will point out that I-71 North in Cincinnati, Ohio, has doubled in lanes over the last few decades and has never been uncongested at rush hour. All we’ve gotten is more northern suburbs while congestion has stayed pretty much the same.
And you want proof that there are people in Cincinnat who live near sewer pumping stations that pay the same sewer rates as the far neighborhoods? In contrast, do you know of any U.S. cities where people pay varying sewer rates based on their distance from sewage pumping or treatment infrastructure?
My overall point is that rather than fixed instructure utility rates being a kindly helping hand to people who happen to live in far flung burbs, the uniform rates and lack of environmental damage taxes on low-density developments has helped hide the costs of low-density living from the average low-density dweller and the populace in general. The result is people thinking that living in the city is a less “green” lifestyle than living in a suburb, when the opposite is true.
And, people think that widening highways can get rid of congestion, sort of the way people think that parking in their downtown would be so much easier if the city would just get rid of all the dang parking meters.
March 12th, 2009 at 12:38 am
Russell,
You can’t relieve congestion unless you know what causes it. You can’t know what causes it without examining the differences between congested areas and uncongested ones. The difference between MSAs with very high congestion, such as Los Angeles (100%) and New York (87%), and MSAs with much lower congestion, such as Denver (15%) and Orlando (5%) is not road pricing. It’s population density and land use. The relationship between congestion and density has been studied academically. And the relationship is that congestion decreases as density decreases. Sprawl reduces congestion. Density increases congestion.
And the claim that congestion cannot be reduced by increasing road capacity is just absurd on its face. Whatever “induced demand” effect there may be from increasing road capacity, demand is ultimately constrained by population. It simply is not true that demand can increase indefinitely to offset any increase in capacity, no matter how large. That doesn’t mean increasing capacity is always the most practical or cost-effective way to address congestion, but it does mean that increasing capacity is one of the ways in which congestion can be reduced.
March 12th, 2009 at 2:38 am
A properly priced road will tell you whether you built a big enough road or need a bigger road.
Let’s say the road is initially built with 3 lanes and a variable peak toll of $5. The tolls collected, it turns out, just cover the cost of construction. What a wonderful equilibrium — drivers are enjoying the precise amount of capacity they are willing to pay for.
So if toll revenue is steady — just paying for construction — do you add more capacity? No, because the little market you’ve created tells you drivers aren’t willing to pay for more capacity; they’re only willing to pay for the capacity they’ve got.
If demand for the road increases and the peak toll rises to $6, then the road is generating extra revenue. Drivers are showing they’re willing to pay for more capacity.
That’s how you should handle induced demand. The driving population — both those using the road and those who would but are deterred by congestion — will pay you money that tells when its time to expand a road. It will be efficient, provided the full costs of the additional capacity (environmental, harm to adjoining neighborhoods and properties, etc.) are factored in.
March 12th, 2009 at 2:42 am
Mixner, have you thought about getting your own blog? I don’t mean it as an insult and am not telling you to go away. It’s just that you are interested in this stuff and spend a lot of time debating it, . . .
March 12th, 2009 at 12:56 pm
AC,
You didn’t answer my question. You claimed that road pricing is the only way to deal with congestion. But the empirical evidence shows that this just isn’t true. Road pricing is rare. Only a very small fraction of road usage is priced. Despite that lack of pricing, some roads are much more congested than others. Some cities are much more congested than others. And this variation in congestion isn’t small, it’s enormous. So factors other than pricing must have a dramatic impact on congestion. The obvious ones being density and land use. Road pricing is just one of a number of policies that can be used to reduce congestion. It’s not a panacea. For most road usage, pricing isn’t yet a realistic option and may never be realistic.
I have no interest in starting a blog.
March 12th, 2009 at 4:50 pm
Just because a road is not congested doesn’t mean it had dealt with congestion, Mixner.
March 15th, 2009 at 12:54 am
AC: Mixner is a troll. Trolls do not have blogs.