Building off of Henry Grabar’s Slate piece yesterday that references our previous work on this, a little more on what we need to know first before we can debate the causes of any shifts that might exist in rural and urban vehicle travel.
1. The headline urban/rural travel figures produced by the FHWA and cited in the Sivak/Schoettle paper mash together three things:
(a) Changes in the number of miles driven by residents of a given “urban” geographic area over time.
(b) The driving behaviors of new “urban” residents settling in newly “urbanized” land on the fringes of metropolitan areas.
(c) The driving behaviors of previously “rural” residents who became “urban” residents only by virtue of having their homes absorbed into an “urban” area as it sprawled.
(I’m using “quotes” here b/c Census definitions of “urban” areas include some suburban and exurban areas that no one would confuse with Midtown Manhattan.)
2. When people hear “urban VMT is increasing,” I suspect that they think first of (a), maybe secondly of (b), and probably not at all about (c).
3. The same is true in reverse of “rural VMT is declining.”
4. Even this breakdown oversimplifies things greatly, as VMT within a given slice of a metropolitan area is not limited to the people who live there. For example, the Austin case we explored a few years ago saw large chunks of I-35 – a main intrastate and interstate travel route in Texas – reclassified from “rural” to “urban.” Any long-haul trucker, salesman from San Antonio, or vacationing road-tripper from Minnesota traveling on that road would have seen their VMT shifted into the “urban” column. But we’ll give that issue a pass for now.
5. Individually, (a), (b), and (c) are all fascinating and raise important research questions:
(a) The first gives us a window into how policy, economic conditions, or individual preferences in a particular place might be affecting travel over time.
(b) The second gives us a window into how land-use change (sprawl) affects vehicle travel.
(c) The third tells us how residents of an area that is “filled in” by “urban” development might behave. For example, residents of a formerly rural area might drive less as destinations move closer to them, or they might drive more because a greater number of destinations are available.
6. The FHWA’s aggregate time series trends for urban and rural travel do not allow us to answer *any* of these questions with any specificity – or even give us enough data to frame up intelligent questions.
7. They *might* do so if the data were clear about when in time the transition in urban boundaries occurred. Then, at least a researcher would be able to isolate the effect of urban boundary changes in time.
8. The Census Bureau makes this clear for population data – every population estimates dataset for urbanized areas is abundantly clear about what set of boundaries it is using. The FHWA doesn’t, and doesn’t even make clear to people using the data that it’s a potential issue. It’s an open invitation to researchers to misinterpret the data
9. If you want to answer any of the relevant questions about these trends, you really need to do one of three things:
(a) Go through the FHWA data state-by-state and urbanized area-by-urbanized area in an effort to find the discontinuities in the data – the years when urbanized area definitions changed. (You could also call up state DOTs and the FHWA to try to find out when they switched from one set of boundaries to another). We did a bit of this here and had to account for it in this report here. Or:
(b) Do GIS analysis of travel data using consistently defined urban boundaries. We did a bit of this here. Or:
(c) Rely on travel survey data.
10. The latest comprehensive, national travel survey data we have (which are from, grr, the 2009 National Household Travel Survey) found that declines in daily travel since 2001 were not all that different between urban and rural households.
11. If you want to believe that there’s been a dramatic per-capita decline in rural household driving since 2000, you either need to a) discount the survey data, or b) surmise that the vast majority of the decline has happened since 2009. (Beyond households, there may have been important shifts in freight travel as well, but FHWA’s data on truck VMT have time-series discontinuity problems of their own.)
12. In short, I’m not convinced that there has even been a significant decline in rural per-capita driving in any meaningful sense, or that the increase in urban driving – in the sense most people would understand the phrase “increase in urban driving” – is as large as it seems from the headline data. It’s possible, but I’m not convinced.
13. If the data don’t allow us to come to agreement on that basic point, it gets very difficult to then engage in an informed conversation about possible causes.
14. And that’s a pity.
15. In the end, this is another example of how we Americans lack a precise and rich language for even talking about urban change; how our existing data sources don’t help us to develop one; and how the unreliability of the existing data sources that we do have vexes us as we try ever so hard to pound them into the square holes we’ve created.