No Light Rail in Vancouver!

Density Increases Congestion

Jan 24

2007

In an earlier post, I mentioned that density increases congestion and was chastised for failing to prove it. So here is the evidence.

Start with data from the 2000 census that compares the percentage of commuters who drive to work with the population density of nearly 400 urbanized areas (areas with more than 50,000 people). If density reduced congestion, then increased densities would greatly reduce auto commuting.

But as the above figure shows, the relationship between density and congestion is weak. Notice that the densest urban area is about seven times denser than the least-dense area, yet the share of auto users in the densest area is at least 90 percent of the share in the least-dense areas.

The figure does show a number of urban areas in which less than 80 percent of people drive to work. But density does not seem to be a factor. In fact, a close look at the data reveals that many of these areas are either university towns, in which many commuters walk or bicycle to work. The rest are areas like New York, which have a huge number of downtown jobs, allowing transit to capture a large share of the commute market.

We can draw a straight line on the figure from the densest to least-dense area and see that very few areas are above that line. So density may slighly reduce per-capita driving. But if it takes seven times the density to reduce per-capita driving by 10 percent, you end up with 530-percent more cars on the road per square mile of land at the denser level.

Nearly all of the claims I hear that density reduces congestion are based on data showing that density reduces per-capita driving. These are two very different things. To reduce congestion, per-capita driving must fall by a greater percentage than the increase in density.

For example, the Transportation and Land Use Coalition, a San Francisco Bay Area group advocating for more land-use planning to reduce driving, says that higher density communities “supports much higher levels of transit.” As evidence, they use the following numbers, which they attribute to a household survey done by the Bay Area Metropolitan Transportation Commission:

The densest neighborhoods generate only about 20 percent as many auto trips per household. So far so good. But let’s add a third column to the table, one that the Transportation & Land Use Coalition neglected to include:

Suddenly, the densest neighborhoods are generating sixteen times as many auto trips per acre as the least-dense neighborhoods. That doesn’t sound so attractive. Unless you have something close to sixteen times as many roads per acre, you are going to have a lot more congestion. Those who argue that density reduces congestion are either deluding themselves or attempting to mislead the public.

Some analysts think that if density gets very high — denser than Manhattan, which is almost ten times denser than the densest urban areas in America — that adding density can reduce per-capita trips enough to actually reduce congestion. But we don’t really have enough places in the country dense enough to measure this, so I am dubious.

Beyond this, correlation does not prove causation, and household surveys such as the one on which the above data are based contain a built-in bias. People who don’t want to drive will tend to locate in pedestrian- or transit-friendly neighborhoods. People who want to drive will tend to locate in auto-friendly neighborhoods. This means that the people answering the surveys are predisposed to drive or not drive. Maybe the main reason there are fewer trips per household in the denser neighborhoods is because people in those neighborhoods don’t want to drive.

If dense neighborhoods have lower per-capita driving because the people who live there want to drive less, then forcing higher densities on people who want to drive may not greatly change their travel habits. Census data show, for example, that people in dense inner cities tend to have far fewer children than people in low-density suburbs. Would putting families with children in dense neighborhoods lead them to drive less? Probably not much.

By the way, density also increases air pollution, mainly because cars pollute more in congestion. Density may or may not increase crime, but it is much more difficult to design crime-free neighborhoods that are dense because the things that do reduce crime — reducing ease of access and increasing ease of identifying who belongs in an area — are more difficult to build into dense neighborhoods. I’ll leave the proofs of these propositions to the readers or to future posts.

10

Trackback  •  Posted in Transportation, Urban areas  

Reprinted from The Antiplanner

Households  Vehicle Trips

    Per Acre   Per Household

       1.4           6.4

       3.6           5.9

       6.7           5.0

     13.5           3.8

     30.6           2.9

   121.9           1.2

Households  Vehicle Trips  Vehicle Trips

    Per Acre   Per Household    Per Acre

       1.4           6.4            9.0

       3.6           5.9           21.2

       6.7           5.0           33.5

      13.5          3.8           51.3

      30.6          2.9           88.7

     121.9         1.2          146.3