Getting the Dependent Variable Right: Trips versus Distance Traveled

There is no question that in the United States driving is declining, and that Vehicle Miles Traveled (VMT) have also declined in absolute and in per capita terms. There are questions as to why, however.  It’s important to understand why and how travel is declining, and we have decades of research that suggests that people consume travel in economically predictable ways by consuming more or less based on income. USPIRG and others are reluctant to view economic reasons as the main cause of the observed decline in distance traveled.  I don’t think distance is a particularly good dependent variable on which to analyze economic factors, though, as annualized distance traveled doesn’t really tell us anything about the economic value of travel. A better metric for understanding the economic factors associated with declining travel is trips, especially trips by purpose. What the data show is not just that people are driving fewer miles, they are also taking fewer trips. 

Personal travel has some economic characteristics that are especially relevant here. Personal travel is a normal good in that as income goes up people will consume more of it. This also means that as income declines people will consume less of it. Conversely, transit, certainly as practiced in the US, has many characteristics of an inferior good where demand declines as income goes up (remember I’m talking about inferior goods as an economic concept. This has nothing to do with the actual quality of transit service.). This is because people replace transit trips with auto trips (or walking trips in certain areas such as parts of Manhattan) as they can afford it.  Now, perhaps since there are lots of new rail systems that were designed specifically to attract people who would otherwise drive we shouldn’t view transit as an inferior good in all situations. I suspect this is true in some cases, but not in most.

The other economic characteristic of personal travel is that it is a derived demand, and distance traveled is a function of the trip demanded. This means that, by and large, people travel because of what they want to do at their destination: work, shop, eat, school, etc. Some of these destinations are close and some are far, but rarely do people decide how far to travel before they decide where to go. As economic conditions for individuals and households change, especially decline, they are likely to make fewer trips out of the home, which will lead to less distance traveled. This is especially true if the areas declining economically are sprawled metros. If trips are declining is sprawling areas then this will lead to unusually large distance traveled declines in surprising areas. 

An an example of the economic value of travel, if a person drives to get a hamburger the value of the trip (utility) is derived from eating the hamburger, not the distance the restaurant is from their house. If someone drives two miles for a hamburger the utility of the trip is nearly identical to driving one mile for the same hamburger, as the marginal cost of the additional mile is small. It certainly isn't the case that the two mile trip has twice the economic value as the one mile trip. 

A more accurate way of looking at the economic influences of personal travel is to look at trips, in part because personal travel is mostly non-work and non-commuting. Looking at data presented in Commuting in America III highlights some of the issues with trips. (The .pdf is secured so copying the tables is more time  consuming than I will take time to do here, but you can read the tables in the report.) Table 1-1 shows the relations between person trips and person miles of travel. Only 18% of distance traveled is due to commuting. Figure 1-2 shows the trips per capita by type for the past few decades (all data from NHTS/NPTS). The number of work trips is fairly stable at just over .5 per capita per day. Personal business and social trips are where the increase in travel happened, where these now consume over three trips person per day, which is about doubles from the late 1970s. The CiAIII report also shows that work trips have declined as a share of overall travel (by trip and total distance) substantially.  According to the 2009 NHTS average annual miles traveled getting to and from work have declined since their 1995 survey (Table 5 below and available at the link above.) The number of commute trips has declined in the same period, too, from 676 per person per year to 541. Moreover, the average distance traveled per trip (for all trips) increased slightly from 11.7 miles to 11.9 between 2001 and 2009.

Table 8 in the 2009 NHTS report shows person trips by household income, and it shows that as income increases the number of trips taken increases (trips as a normal good). Table 11 shows miles per trip type, and it shows that the total number of commute trips per person has declined along with miles traveled. In addition, all trip types have declined in number per day. Looking at Table 6, the average vehicle trip length is largely the same between 2001 and 2009. (Check the link for the tables as I don't want to make this post longer than it already is.) Taken all together, it seems likely that the decline in distance traveled is due to a decline in the number of trips taken. I argue that this is an important distinction when we think about policy.

If we think that people are doing all the travel that they used to but are now traveling shorter distances, then the natural policy response is to physically rearrange our cities. I suspect this is a reason that the VMT declines are so compelling to urbanists who prefer compact development, as their desired urban form happens to be useful policy in this case. Yet the decline is distance traveled is less associated with compact physical structure of cities and more associated with fewer trips taken. One way we know this is that metro areas were continuing to sprawl as distance declined.  If anything, the decline is distance traveled is associated with the suburbanization of employment, shopping, entertainment and recreation. There is evidence that this is the case. Ultimately, while some downtowns and town centers are doing very well at attracting new residents, the total number of people moving to these locations is fairly small (maybe a couple of hundred thousand across the country so far, and that’s a generous estimate) and not enough to offset the decline in distance driven. Nor are people substituting transit for driving in any great number. In some cases, it is absolutely happening, but most of the driving that is no longer happening cannot be reasonably substituted to transit.

If, instead, we consider that people are not traveling as far because they are not traveling as often, then we may be looking at behavioral responses to declining utility of travel. Why are people not taking trips they used to?

Here are some reasons, all of which have economic justifications related to the economic characteristics of personal travel mentioned above. These are also global conditions, which fit with observed declined in travel in developed countries around the world.

•       Household structure has changed. People are putting off having kids, and kids are associated with a lot of travel.
•        Families with children have fewer children. This is especially true for wealthier households.
•        Young people are broke and unemployed with lousy job prospects.
•        Cars are increasing in costs faster than inflation.
•        Consumption trips are substituted for delivery.
•        The relationship between workers and office/factory is changing.
•        There is an upper limit constraint on how much travel people will consume in any day.

There are good reasons to believe that automobility hasreached “peak utility,” but this doesn’t necessarily mean that other modes will be the beneficiaries. It may just mean that people don’t travel as often.

A Rural Explanation for VMT Decline?

Vehicle miles traveled (VMT) have been declining in the US, Japan and Europe. There has been a lot of interest in what is causing the US shift, and much of the interest is in whether shifting demographic preferences have caused the decline, especially among young people. Less attention has been given to geographic differences, specifically urban and rural distinctions. The Brookings Institute made note of these distinction in their 2008 report The Road...Less Traveled. Yet it seems this distinction may deserve more attention than it is getting. The figure below shows the total VMT for the US plus for urban and rural travel up through the year 2011.

You can see the overall decline in VMT that starts in 2007 after the peak of over three trillion miles. The green line shows urban travel, which stopped increasing and leveled off, and the red is rural travel, which started to decline in 2002. At a glance it seems that much of the decline in overall VMT can be attributed to declines the rural road networks. This does not preclude demographic shifts, but people haven't been moving as much in the US and there hasn't been a dramatic rural to urban migration starting in 2002. Potentially some of the rural decline could be from land getting reclassified from rural to urban, but I don't expect this would be a big effect.

Here is a table that shows the data from the above figure along with the relative changes from peak VMT for rural and urban areas. Where rural VMT has dropped 13% from the 2002 peak, urban VMT has dropped less than one percent from the peak in 2007. That the relative share of VMT has increased in urban areas is just an artifact of the rural decline. Travel reductions are not evenly distributed.

Rural and Urban Vehicle Miles Traveled in US 2002-2011
Rural		Urban		Total		% Rural		% of Rural Peak		% Urban		% of Urban Peak
1.1274		1.7281		2.856		39.48%		100.00%		60.52%		86.64%
1.0844		1.8058		2.890		37.52%		96.19%		62.48%		90.54%
1.0684		1.8964		2.965		36.04%		94.77%		63.96%		95.08%
1.0324		1.9757		3.008		34.32%		91.57%		65.68%		99.06%
1.0371		1.9772		3.014		34.41%		91.99%		65.59%		99.13%
1.0353		1.9945		3.030		34.17%		91.83%		65.83%		100.00%
0.990418		1.9831		2.974		33.31%		87.85%		66.69%		99.43%
0.98218		1.9746		2.957		33.22%		87.12%		66.78%		99.00%
0.984148		1.9824		2.967		33.17%		87.29%		66.83%		99.39%
Miles in trillions

The decline in rural VMT is partly because rural areas are associated with so much more travel than urban areas. States that have higher shares of urbanization has lower VMT per capita than the national average. Using data from the FHWA I calculated the correlation between the percent of the population that is urban and VMT per capita at -.58, which is a fairly strong association between increased urban population and lower VMT. 

Certainly there are changes afoot in the US economy and transportation. This post is not intended to make any dramatic claims about VMT declines. However, these data suggest that the bulk of decline is from rural reductions, not urban reductions. Of course, we need more research about this.

Do Local Policies Explain the Decline in Driving?

Driving is declining. Reports like this recent one by USPIRG claim that "The Driving Boom—a six decade-long period of steady increases in per-capita driving in the United States—is over." The US Department of Transportation reported that these trends are continuing, and there was an additional decline of 1.5% in vehicle miles traveled year over year in March. While it is clear that driving is declining, it is less clear why this is happening. Common ideas as to why personal travel is declining in the US include land use changes, transit investment, smart phones and environmental concerns. These may or may not be accurate ideas (I'm skeptical.). This recent report from UCLA suggests that economic factors play a larger role in shifting driving behaviors for young adults.

Only focusing on the decline in driving in the US paints an incomplete picture of associations and policy interventions that may (or may not) have encouraged the shift. I am not convinced that local policies have much to do with the decline in driving in large part because the decline in driving is global (at least in developed nations). Here are data from the UK showing the same kind of decline, and here is a story about research published a couple of years ago covering eight countries. Car companies in Japan have been worried about young people turning away from cars for years, and youth attitudes showing declining interest in driving have been fairly consistent over the past few years. Since these trends are global we should first consider globally shared conditions for explanations as to why travel is going down. Historically high rates of unemployment among young adults is good place to start. Local policy is too heterogeneous to account for overall decline.

Driving May Be Down, But Drive-Thrus Are Booming and So Is Parking

According to the LA Times America's drive-thrus logged 12.4 billion trips last year, which is a 2% increase over previous years. So whatever the reason that US drivers are traveling fewer miles in their cars it seems that we can assume they aren't foregoing their fast food rips. (And remember that it is the middle class that uses fast food restaurants more than anyone else.) It boggles my mind that even though 70% of fast food sales are from the drive-thrus there are still minimum parking requirements of about one space per 200-300 square feet of restaurant space. Yet here (found with a very quick Google search) is a Michigan McDonald's with a drive-thru that seeks to build 45 spaces when 27 are required. Sheesh. Whatever you think of the health effects from fast food, the land use and transportation implications are extremely challenging for urbanism.

Are VMT a predictor of a recession?

Last year when vehicle miles traveled (VMT) were plummeting, I suspected that the spike in gas prices was less of a driving force for declining VMT than rapidly shrinking housing activity. Part of my evidence for this was based on the uneven increase in transit use. If gas prices were shifting people away from driving but the demand for travel remained constant then most of the shift should be captured by transit. The rest of the shift would go towards car pools, walking, biking and telecommuting. But the case was that overall travel was declining, and I thought that this was a leading indicator of a weakening economy.

I never got around to making a chart showing this, so fortunately Calculated Risk has finally produced one. It's pretty clear that a rapid and large decline in VMT growth occurs right before a recession. Once VMT growth hits about one percent or less year over year, the economy goes bad. Obviously this is a simple apparent correlation and does not suggest causality. Since travel is largely a derived demand, economic performance is a major predictor of VMT. But in the future when we owonder hos the economy is doing we should look to VMT as a proxy measure of health. When VMT declines rapidly we should worry about the overall health of the economy rather than patting ourselves on the back for increased light rail transit boardings.

The chart is from calculated risk.