Friday, March 31, 2006

"If it made sense, that would be a very powerful idea," or Land as a Transportation Planning Tool

Let’s get this much straight at the beginning: Land is a critical transportation planning element and you’d have to be oblivious to the world around you to even imagine otherwise. Plus, other physical planning strategies to deal with traffic don’t seem to work. People keep driving more and more, everywhere.

That said, it’s fair to ask if it is a useful tool.

The sheer quantity of planning research on land use/transportation linkages in the last 15 years is nothing if not high and wide. Moreover, renewed interest in the public health potential of activity-promoting urban design greatly accelerated volume on both counts. The Robert Woods Johnson Foundation and the Centers for Disease Control ramped up their spending on this work, funneling in millions of dollars, as did many state and federal government agencies, in large part aimed at advising interventions to address the “epidemic of obesity,” often mentioned as a fast approaching runner-up to smoking as the leading discretionary cause of death in the U.S. Susan Handy wrote a solid literature survey recently, commissioned by the NRC committee assigned to the topic, with a bibliography of dozens of recent articles.

It’s usually around this point where some overbearing assistant professor asks something like, “And what has all this research taught us about what to do? Not very freaking much.” That’s not especially helpful, at least not when your city councilman wants to act. Can research be helpful?

The trouble with traffic
As taught in both elementary and planning school, first define the problems at hand. Consider two: Time stuck in traffic and lack of exercise. The first is costly in itself, as time truly is money, as well as associated with an induced loss of “sense of community” and other measures of social capital, air pollution, accidents, and so on. The second leads to premature death. (I’m convinced by the evidence that modest levels of activity have significant health benefits.)

The potential of land
Where does land fit in? One, it regularly separates us from where we want to be. Two, the road network and transportation system are components of urban form, including circulation patterns, traffic bottlenecks, sidewalk provision, the street hierarchy, and so on. Three, the specific pattern and details of a given built environment may introduce further delays, or interact with distances, or somehow favor some travel means over others via mixed land uses, development densities, and more. I could go on.

Thinking about this narrowly, land use can thus influence the costs and time associated with any particular trip. In turn, it can also influence the mode chosen, by facilitating or discouraging (lowering or raising the cost or quality) of alternatives. (Do densities make transit viable, and how long would a bus take for a particular route compared to walking, the train, or a private car?) In tandem, land use can help determine what a given trip can accomplish. Mixed land uses facilitate multipurpose trips, for example. The efficient configuration of shopping, employment, and recreational destinations can promote trip chains or tours that effectively reduce the travel times required to accomplish a list of tasks.

So it follows that purposeful land changes (a) might get us closer to where we want to be and that this (b) might get us out of our cars, which in turn would clearly lengthen lives. These are big deals, so it is no wonder why there is so much interest in land as a transportation planning tool.

Research questions
I go back to the central questions of our book, TBD: Will it work, can it be implemented, and is it a good idea? Our answers were: It depends, it depends, and it really depends.

These were very helpful results, I think, as the choices to that point were more on the order of yes, yes, yes or no, no, no, without much guidance about how to negotiate the terrain between. TBD explains how and why but runs $74. The 30 minute exam version would look something like this:

1. Will it work?
I argued above that land use planning could reduce travel costs and that this in turn may reduce traffic times and congestion, and promote physical activity. The key word is could. Do we know exactly when it would do that? Is this a simple and fairly robust relationship (e.g., sweet talking a camel into going forward) or a complex and hard to predict one (e.g., sweet talking a teenager into becoming a responsible adult)?

Until the late 1990s, most planning studies simply regressed trip lengths on different measures of land use. The most popular such measures were of mixed land uses, higher densities, and more open circulation street patterns, all of which reduce trip length. The best work developed better measures at finer levels of detail but, oversimplifying a bit, they essentially tested – and confirmed – whether trips became shorter if the land use shortened trip length.

Does this mean planners should promote compact development? If the goal is only to shorten the length of a given trip, absolutely. Will car use then decline and pedestrian travel and transit demand rise? If it were only that easy.

The rub is one of two main points in a 10-year-old JAPA article. Demand curves are downward sloping, so if trips get shorter, people will take more trips. (Thankfully, the editors told me to lose the appendix formally demonstrating this result, which I then turned into a JPER article. I was up for tenure and wasting no part of the animal.)

This is a fairly unambiguous result, applying virtually everywhere. Instead of just shorter trips, you’ve got more trips in less space. Not good for congestion, parking, or cold starts. VMT could even rise. The best you can hope for is that some of these trips shifted to other modes. That could clearly happen. In any event, we no longer have a sure fire cure but, instead, what social scientists call “an empirical question.”

Good, that gives something to do to those of us looking for work. But it has to be done convincingly or the results are idle chatter. In particular, if the built environment influences behavior by affecting marginal incentives, then by golly shouldn’t the statistical strategy explicitly account for that? You’re right, of course it should. Yet this remains rarer than you might like to think. Indeed, I have heard people – JAPA referees even – argue that worrying about the merits of one empirical strategy over another, when so much is at stake, is nitpicking and obstructionist. I suppose they are saying, in their own way, that when the answer is pretty obvious you shouldn’t worry about a few loose ends. Fair enough. But in this case, there is more under debate than extraneous details. Plus, these are often easy enough to properly analyze, so why not do so? Stand up and be a proud planning researcher whom other disciplines will take seriously. ¡Si se puede!

Here is the bottom line: Land use changes can modify both the physical parameters of travel (e.g., trip length) as well as underlying, elemental travel incentives (e.g., unit travel costs). Ignoring or mischaracterizing the latter is simply not a risk worth taking, not the least since it is unnecessary.

My own view (a qualification I must be adding for dramatic effect only, since I have no editors or referees looking over my shoulder here), is that barely a fraction of the more recent research on built environment determinants of physical activity are informed by this logic. Any positive correlation between a feature of the physical city and a measure of walking or biking seems to be publishable as evidence that that more of that feature would save lives. How that feature translates into marginal incentives, or the importance of controlling for confounding effects (such as self-selection), is often absent. So long as we are simply getting a handle on correlations, ok, but this is no basis for policy recommendations.

An example
Say we want people to walk more (so they’ll live longer). Making trips shorter would increase trip taking but the biggest stimulant for walking is probably the switch from other modes by shortening trips to make walking more feasible or safer. The "pedestrian friendliness" of a route certainly matters, but how exactly? How it affects the effort or pleasure or feasibility of a trip at the margin, relative to other modes and route characteristics, should be explicitly considered.

This is supported by several studies that explicitly account for such substitution effects, and sample bias, such as a series of papers by Greenwald, Boarnet and Greenwald, Boarnet, Fulton, and Nguyen, and others. With other studies, that do not model substitution parameters or selection bias, it is harder to generalize outside the sample under examination.

2. Can it be implemented?

Revisiting Levine, his critique has two parts. One, evidence of self-selection is not evidence that providing more compact neighborhoods has no behavioral effect. Two, higher density and thus more walkable, transit-friendly neighborhoods are undersupplied. So even if we got the prices right, there is unmet demand.

The second point is an empirical question, for which his book has summarized considerable supporting evidence. TBD also tests this question in a limited way, with respect to the supply of transit-based housing, with similar results.

I repeat, however, that accepting this point does not, in itself, indicate that ignoring how the built environment affects travel incentives is ok. Indeed, he sometimes seems to be saying that the entire travel benefit of providing more compact neighborhoods may be the so-called self-selection effect. Say that is true. What does it then mean if higher densities do not change mode shares or reduce congestion? So far as I can see, it does not mean that land use is a reliable transportation planning tool.

3. Is it a good idea?
So we are left with more questions than hard answers about how to apply land use in the fight against traffic problems. It clearly matters. How and when it matters is less clear, though with more elbow grease we are deepening our knowledge base, particularly for specific settings and details.

If we knew how to reduce car use and promote less or other means of travel, whether this is advisable is even harder to say. Sometimes probably yes, other times perhaps no.

There is no secret to reducing mobility through other mechanisms. We could stop building roads, raise the price of gasoline, etc. The tradeoff is that discouraging car use is akin to discouraging international trade. There are benefits to restricting mobility and interaction, to be sure, but what of the costs? This question arises where we can reliably use land use toward that end as well. I leave that for another time.

Is careful empirical research helpful?
Yes, for at least 3 reasons. 1, it delivers credible evidence one way or the other, indicating what we do and don’t know. 2, it tells us how to learn more. 3, it therefore clarifies the facts, which we need to make value judgments. Without the facts, we are not helpful to anyone, no matter that our hearts are in the right place.

No comments:


My blog has moved!

You should be automatically redirected in 6 seconds. If not, visit
and update your bookmarks.