The four biggest myths about induced demand

Two weeks ago, I laid out the economic argument for induced demand: the idea that building more roads does not reduce congestion. It is a simple model that uses concepts from Economics 101 to explain the relationship between road construction and driving behavior.

Even so, this idea, like many associated with new urbanism, challenges the status quo. As such, there’s pushback. To ensure that no falsehoods go unchallenged, I decided to examine the claims in two articles that seek to discredit induced demand as a property. The first is a blog post from the Cato Institute written in response to a Wired article on the subject published last month, and the second is a Weekly Standard story written three years ago (which is still fresh in urban planning time). Here are the four most prominent false assertions upon which the articles rely.

1. Since roadway capacity is not the only factor affecting driving, induced demand is a flawed model.

This misrepresentation was trotted out by the Cato Institute, which attempted to discredit the academic research of Gilles Duranton of the University of Pennsylvania and Matthew A. Turner of the University of Toronto, who measured the elasticity of demand of vehicle miles traveled.*

On average, driving grew more than twice as fast as lane miles. But in Boston between 1983 and 1993, freeway capacities grew by less than 1 percent, while driving grew by more than 35 percent. In Madison, capacities grew by 35 percent, while driving grew by less than 20 percent. The wide range in differences between urban areas suggests that, not only are Duranton & Turner’s elasticities wrong, their standard errors are far too low.

Duranton & Turner’s headline finding was that the elasticity of demand in the transportation market is 1, according to roadway data from 1980 to 2000. In other words, holding other factors constant, a 20 percent increase in roadway miles elicits a 20 percent increase in vehicle miles traveled. “We found that there’s this perfect one-to-one relationship,” Turner told Wired.

Cato fails to account the other variables that affect driving patterns like geography, population growth, and socioeconomic characteristics that Duranton & Turner specifically control for. Simply noting that all cities’ freeway capacities and driving patterns don’t fluctuate in lock step does not show anything.

Duranton & Turner sought to find the relationship between two variables alone and found a striking relationship. We live in the real world, and there are other factors that affect people’s behavior.

2. Cities that have invested in public interstates have seen long-term reduction in congestion.

The Weekly Standard blithely throws out this claim without qualifying it in any way:

The Texas Transportation Institute’s annual Mobility Report, for instance, demonstrates an uncanny correlation between capacity and traffic congestion: Areas that add capacity tend to have lower levels of congestion.

First off, that’s not what the the authors of the Texas Transportation Institute’s Mobility Report found. Instead, they wrote that “additional roadways reduce the rate of congestion increase,” which is a substantively different assertion.

Additionally, their analysis is based on the assumption that roadway growth (supply) and vehicle miles traveled (quantity demanded) are independent of each other. While there are certainly other factors involved, the built environment contributes significantly to people’s behavior. Ignoring this fact is tantamount to building a new road, observing an increase in vehicle miles traveled, then assuming it would have happened anyway. This methodology leads to a skewed result, which isn’t matched by other studies.

The most robust study on the relationship between congestion and roadway growth comes from the Victoria Transport Policy Institute, which found that “Traffic congestion tends to maintain equilibrium. Congestion reaches a point at which it constrains further growth in peak-period trips. If road capacity increases, the number of peak-period trips also increases until congestion again limits further traffic growth.”

Plenty of academically-minded people before me have established the economic model. For one, Douglass B. Lee Jr. at the World Bank provides a more rigorous explanation. A meta-analysis of induced demand studies by Robert Cervero in 2001 found strong evidence of the existence of the phenomena, though different researchers have established different elasticity quotients. Recently, Duranton & Turner derived an elasticity of 1 with a very low standard error.

(The Texas Transportation Institute study has several other problems that Tanya Snyder at Streetsblog USA and Todd Litman at the Transport Policy Institute can address more thoroughly than I can.)

3. Automotive transportation is the most efficient way of moving people around a city.

This contention isn’t really even a myth—it’s a fabrication. It has no basis in reality. This point was appended to the end of the Weekly Standard article:

A metropolitan area typically has about half as many jobs as people. But, because of geographical constraints, not every job is accessible to every person. Highways are, far and away, the most efficient way of delivering people to a job.

In most American cities, auto transportation is the best readily-available way to transport people because there are no other options. That does not in any way imply that it’s the most efficient way to organize a city. On the contrary, car dependence is both inefficient and wasteful:

• University of Michigan study: “Overall, in 2010, BTU per person mile was 4,218 for driving versus 2,691 for flying. Other modes of transportation: Amtrak trains (1,668), motorcycles (2,675) and transit buses (3,347).”
• Portland’s dense development patterns yields $2.6 billion in yearly savings, which amounts to a 3.0 percent income bump relative to the average citizen of the United States.
• Automobile congestion as a result of publicly-subsidized highways costs Americans at least $45 billion every year.

4. Vehicle miles traveled isn’t an important metric.

This is a confused contention that doesn’t hold up to any scrutiny. For some reason, we shouldn’t be focusing on vehicle miles traveled as a metric because… we don’t like doing it? Again, from TWS:

[Principal at Demographia Wendell Cox] maintains that the entire framing of the issue is faulty: “Latent demand” for a highway, he notes, isn’t actually a desire to drive on that stretch of road. People only want the road as a means to an end. “Transportation is not a primary activity,” Cox explains. “There is no ‘love affair’ with the automobile. Driving is not something we would choose to do.” […] In other words, a metric like “vehicle-miles traveled” is only superficially important. 

Plenty of economic goods are means to an end. No client wants to pay up to mount a legal defense, but they do it anyway because they have to. Just because it’s a means to an end doesn’t mean we have to spend the money for it. With dense development and healthy public transit, families are able to spend less time in traffic and fewer dollars on gas without sacrificing mobility. Maximizing vehicle miles traveled should not be anyone’s objective.

Induced demand is an economic property with solid evidence

The key insight from the market model is that  increasing roadway capacity will only make sprawl worse and won’t fight congestion. While car dependence hurts public health and wastes money, this economic principle does not imply that all highway construction is misguided. All planning is local. (Like politics.) There are plenty of good highway projects, but they must be balanced with investment in transit so that our cities can be strong, diverse communities where having a car isn’t a prerequisite for full citizenship.

End note

For a more intuitive explanation of induced demand, see this insightful post from Greater Greater Washington on how building public roads to fill “latent demand” is like putting out more and more free hamburgers to feed people.

*Elasticity of demand is the percent change in quantity divided percent change in price that measures responsive consumers are to changing their behavior given a price increase. This quantity can be visualized by the slope of the demand line.

Photo: Mark Strozier via Flickr