►◄ Reverse Zone
 

Home

About
Reverse Zone, weblog on urban planning, sustainability, and technology.

Martin Laplante

Subscribe
to an RSS feed of this weblog.

Links
A few favourite links.

Recent posts

 2009/01
 2008
 2007
 2006
 2005
 Complete List of Posts

Top Posts

Add to Technorati Favorites

Real Estate Top Blogs

Sustainability Web Ring 
control panel

     
Wed, 06 Sep 2006

Is High Density the Cure for Transportation?

It's about time I talk about my paper in last month's Ontario Planning Journal, Optimizing TOD Housing Mix and Density before it disappears from the internet.

In a nutshell, it explains that higher density in the form of houses reduces the total driving done within a city, but high density in the form of apartments does not. The article on the web has been significantly edited by the journal to remove the diagram and the references, in accordance with the style guide of that section of the publication. Here is a longer version.

I got interested in the topic of what is the ideal residential density for Transit Oriented Development (TOD) when I read something written by Peter Calthorpe (I can't find it any more) that proposed some very specific percentages of different housing types in different types of TOD projects, that included a lot of relatively low density forms. Contrast that with the Official Plans in Ontario, in places like the Golden Horseshoe and Ottawa, where density is to be as high as possible. A great deal of those plans are based on having nothing but high-density apartment buildings near where there is transit.

These are two very different approaches. What evidence do either of them have to support this. In Ontario I talked to a number of planners and read a lot of documents. It turns out that there is no real evidence other than extrapolation. So I put together a very simple model based on what I knew about vkt of different groups at different densities and distance from transit or from downtown. All of it was very orthodox numbers, and it showed first of all that there was an optimal density, a minimum above and below which total driving increased, and how to calculated it with really quite simple easily available data.

The paper was rejected. Apparently it was based on unprovable assumptions. Essentially the model was based on the self-selection bias. Build at higher density and the mix of households will tend toward smaller ones. But eventually, the smaller households are on average closer to transit and the larger ones further. The larger ones then drive more. But how do I know that households will move and how do I know which will change their driving behaviour and by how much?

So I went back to the drawing board and took a detailed look at the data released by the 2001 National Household Travel Survey (NHTS) in the U.S. They have a wealth of data and make it available. Rather than doing it group by group, and using statistics to see where people would eventually move to in response to development, I did the analysis based on the housing form, much more directly measurable. What I found was so surprising that I spent weeks going carefully through the weighting factors, sampling errors, and non-respondent corrections to make sure that it was not just a statistical anomaly.

The graph shows the story: for people living in single houses and townhouses, the relationship between density and total amount driven is pretty well according to common knowledge: if you live in higher density you drive less. But for people in apartments, the story is very different. First, in medium and high densities apartment dwellers drive more than house dwellers. Second, increasing the density has less effect on apartment dwellers than on house dwellers. Third, apartment dwellers drive much less in low density.

Also, by trying out the density graph with density measured at different levels (density of your block vs density of your neighbourhood vs density of your area) the effect is very local: block-level density is a better predictor of your car use than neighbourhood density.

For several different variations on types of apartments, the results are similar, and most of the graphs are much worse than this one; apartment dwellers actually drive more, not less, at higher density. But those conclusions are more difficult to be confident about, especially since they are more controversial. Some housing forms are rarer at low densities, so the sample sizes get pretty small for some points.

The results are actually quite shocking and show that not only does apartment-based high density do nothing to reduce car use, it actually has the opposite effect. High density does little or nothing to reduce car use of apartment dwellers. Density does, however, improve the picture when it is applied to houses and duplexes.

So if the objective of a plan is to reduce the total number of vehicles on the roads, concentrate on houses and townhouses and put them in as high a density as you can. As for apartments, it doesn't seem to matter much whether they are in high or low density. You don't get much reduction in vehicle use by putting them in high density.

So why do individual TOD densification projects show good results? This is where self-selection comes in. If the major advantage of your new apartments is that they are near transit, then those people who would use transit anyway tend to move there. That doesn't necessarily mean that there are more people using transit, it means there are more people using transit right in the vicinity. If they would have used transit anyway, then you have no new transit riders. But now you may have one less transit rider living further from transit, and the household replacing them further from transit may well now switch to cars in a big way.

This may be one of the lessons to be learned from Portland and other cities. You can have a lot of individual localized success stories, but you have to look at the big picture. If your overall transit ridership does not grow and your total vehicle distance driven does not fall you have not made any progress.

Tags:

[] permanent link Comments: 2