Pushing high density living may seem like a good way to get people out of their cars—saving them money, curbing emissions, and reducing oil dependence—but densification may not be a silver bullet, according to one recent study. The authors dug into the National Household Transportation Survey to examine per household vehicle ownership rates, vehicle miles traveled (VMT), and fuel consumption. While the results are by no means comprehensive or conclusive, they suggest that only the steepest increases in density could reduce car usage.
Despite a correlation between density and car usage, other factors seem to play more important roles. Density is responsible for a fraction of annual VMT; increasing density by 1,000 housing units per square mile—a titanic leap, given that the average household is 2.6 people—reduces VMT by just 1,171 miles, all else being equal. Since that the average one-driver household in the study tacks on 10,100 miles per year, that represents just over an 11 percent drop in annual mileage.
If you look at the numbers another way, the case for density reducing car usage looks even more tenuous. VMT only really declines substantially at the highest housing density—over 5,000 units per square mile, or about the same as Chicago. To halve VMT of the highest mileage households, you would need to increase housing density in those areas by 20- to 100- fold.
The inflexibility of our automobile usage boils down to a few factors, with work being the most important. The more workers in a household, the more drivers, and the more drivers, the more miles. A one-driver household, as noted above, tallies 10,100 miles per year; a two-driver household racks up 18,800 miles; three drivers, 33,900; four drivers, 47,700.¹ We are, by and large, beholden to our cars because we are beholden to our jobs. After that, driving increases as a result of income (richer people drive more), number of children (more and larger cars), education (higher education means more cars), and people’s life stage (households with older children have more cars).
While higher housing density doesn’t seem to reduce VMT, it does drive down fuel consumption. Households in the 50 to 250 houses per square mile range use 1,650 gallons of fuel annually, the most of any group. Every other group uses far less fuel. In the big cities, fuel usage drops to 690 gallons per household per year.² The reason? People with the space to use pickup trucks, SUVs, and vans tend to buy them more than people who live and drive on tighter city streets—they typically drive smaller, more fuel efficient vehicles. Yet this trend could be changing as we speak. Small car purchases have been increasing across the country, and anecdotally at least, I can confirm that large pickup trucks are harder than ever to sell these days.
One of the main arguments behind higher density living is that it will reduce our carbon footprint. While density may be a better long term solution, right now the most expeditious approach is to increase fuel economy. Rebuilding neighborhoods will take decades. In that time, most people will buy at least a handful of new cars, primarily for commuting to work. It would be great if everyone had access to mass transit, but for many, mass transit isn’t just a poor option, it isn’t an option at all. Those who do travel by bus or train today may only be a job change away from having to drive. Modern life demands mobility, and few things are better at providing that than the automobile.
—
¹ The increase from one to two drivers probably reflects some combining of trips by couples or roommates. The sharp increase from two to three drivers is probably the result of a family’s children driving to school or work.↩
² The lone outlier is areas below 50 houses per square mile, where households use 1,200 gallons per year. They probably have fewer nearby destinations, and so stay home more often.↩
Source:
Brownstone, D., & Golob, T. (2009). The impact of residential density on vehicle usage and energy consumption Journal of Urban Economics, 65 (1), 91-98 DOI: 10.1016/j.jue.2008.09.002
Photo by dsearls.
Related posts:
Tell me how much you drive, and I’ll tell you where you live
Urbanites leave the car behind, but not as often as might think
I think you’re missing the bigger point here. Sure, VMT is inflexible at many housing densities, but your charts make it obvious that it is very flexible at URBAN densities. It isn’t enough to glibly note that 2.6 people * 1000 hh/mile is a lot of people (even if you could assume that hh size is evenly distributed, and you certainly cannot). You ought to describe actual densities experienced in cities.
So, look at this (http://bit.ly/oIY07n) map of house hold densities for Seattle, noting that there are 640 acres per mile. Only the dark blue areas fall below 1280 hh/mile. Since the large blue area in the middle is a port and related industrial uses, nearly all of Seattle is more than 1280 hh/mile, and most is more than 3200 hh/mile (yellow). Central Seattle falls well above 5000 hh/mile, with a maximum density of over 87,000 hh/mile. So maybe 1000 hh/mile isn’t such a huge leap after all.
And with all due respect to multivarible regression, even your charts show a very clear decline in VMT from 1000 to 3000 to 5000 hh/mile. So the real story here is that, while VMT doesn’t vary with HH density at low densities, within cities it matters a great deal. From your chart, it looks like areas with >5000 hh/mile drive 40% less than those with 1000-3000 hh/mile. That’s pretty significant to how people live in cities.
Finally, the authors of the paper themselves make a biased interpretation of their results (which no doubt led to your mistaken conclusions). They say, “Unfortunately for those wishing to use land use planning to control residential vehicle use, it is very difficult to increase the density of an established urban area by 40%….These data show that only 30 out of 456 cities increased population density more than 40% between 1950 and 1990.”
However, it isn’t necessary to increase the density of the whole city by 40%. What is necessary is to increase the density at which the average family lives by 40%, and that is much easier. For example, my courtyard apartment building has 14 units in the same area of land that might hold one or two single family homes. So, if I move from a single family home into an apartment, I have increased the density at which I live by at least 700%.
Methinks these authors have an axe to grind.
Excellent points, Patrick. Still, I think the authors make some valid points, for a few reasons. First, and probably most importantly, employment has a large impact on whether people can use mass transit to commute or not. I have been lucky in that I’ve been able to walk or take the train to work for the last seven years. People may buy a house, condo, or apartment with the good intention of being close to their jobs or within range of good mass transit options, but they may not have that job forever. Should they lose it and be unable (or unwilling) to sell their house, they may be stuck with a driving commute. This problem is only compounded in multiple earner households, as Jason rightly notes in the comment below.
The point is not to say that density is a bad thing—there are a lot of positive sides to higher density living. The point is that density alone won’t solve some of our problems—like traffic or carbon emissions or oil dependence, at least not on time scales that matter for those things. Climate change, for example, needs to be tackled soon and can’t wait for housing stock to turn over or the real estate market to rebound. Saying that higher density living will help climate change isn’t an inaccurate, it’s just an approach that won’t make much of a difference in the near term.
Also, one final point: You mention that increasing population density by 40% should be measured on a city-wide basis, which if I’m not mistaken is how the study they cited measured it. Adding apartment buildings increases local density greatly, but the general pattern has not been toward increasing density.
The authors of the study (and, as I interpreted it, you) seem to be trying to diminish the value of density in the minds of planners and policy makers. OF COURSE it isn’t a silver bullet – nothing is a silver bullet – but that doesn’t make it less valuable. Things like job location matter, but usually people who live in higher density neighborhoods drive less (as shown in your above graph).
To respond to your final point – you misunderstood. I said that population density DOES NOT need to be increased on a city-wide basis. What matters is the household average density for the city, not the area average density. Consider a city that has 100 households and 100 acres, and 99 of them live in 1 acre but 1 of them lives on the remaining 99 acres. The area average density is 1 household per acre, but the household average density is 98 households per acre. And what matters is the household average density, because that is where driving decisions are made. How the authors measured it in their conclusion was irrelevant to the conclusions that their trying to make.
Another example: If my city has two single family houses, on one acre, and I tear down one house and build 10 units in an apartment building, my area average population density has increased from 2 hh/acre to 11 hh/acre, a 550% increase. But my population average density has increased from 2 hh/acre to 18.36 hh/acre, a 918% increase, since my apartment dwellers live at 20 hh/acre. So, the point is, increase the population density in a small part of the city can have outsized effects on the density at which most people live. And since density makes a difference, it is worthwhile for planners and policymakers to encourage high density living.
Ugh, “they’re” not “their”
Job diversity may play a role here as well. As the job market becomes more and more specialized it seems less likely that a job that fits your profession will be nearby. This is made more difficult when you have multiple earner households.
Well it appears to me…. beyond the numbers the issue that is stated, Drive a lot? Housing Density may not be to blame, that the blame actually lies in land use policies. Specifically those that are not tied to transportation and housing needs. People are driving whether they live in dense environments or rural ones, to satisfy their needs. If we have better land use and transportation policies that are tied to specific master plans and vision from the community. We can allow opportunities for the market to excel while it satisfies the needs of the surrounding residents, eliminating their need to drive for goods and services. Many times I experience this as politicians and or local officials trying to control or direct the market versus providing quality living spaces for people and places for the market to flourish. You can nudge it (the market) but you cant dictate to it.
It is a difficult task to provide for all the needs and wants of a community when the automobile opens up many opportunities to people. Cars do allow you to cover a pretty wide area, quickly increasing choices and the variety in goods and services. It is a difficult task to change that. It is hard to create a small area with large variety (topography itself can cause this. It is hard to get from the beach to the mountains.) or to create mass transit systems that can cover large areas cheaply and quickly. As for master plans and visions, the plans established usually lack the flexibility to deal with future changes and the vision of a community is often not fully shared by the community.
I can’t believe that mixed use has not come up. I’ve never heard a planner say that higher density living will reduce congestion by itself. Only if it is within the context of a walkable, transit-supported mixed-use where “choices” (jobs, shopping, dining, entertainment, schools, recreation, etc.) are located nearby. Arguing a point that has never been made is well…”pointless”.
First — have enjoyed discovering and reading your blog.
As far as the study, I’ll venture to guess that as densities increase, yet remain suburban (as all of the above categories are up to and including 5,000 units/sq. mi.), people will compensate somewhat for shorter driving trips by taking more of them. This is a better outcome in terms of both economic value per mile driven and general quality of life (many more close destinations for less driving), but that’s not visible by looking solely at total miles driven.
To achieve truly walkable urbanism, much higher densities than anything shown here will be needed (10,000 units/sq. mi. plus). I agree that there’s no reason not to promote the construction of neighborhoods like this over the long term while seeking to improve fuel economy in the shorter term, as long as the underlying goal remains reducing dependency on the car, rather than helping people to maintain their driving habits.
A gross density of 5,000 per square mile equates to an approximate net density of 7 dwellings per acre. That’s the density at which frequent transit service just begins to be viable. A bus every half hour, for example. So this study is only looking at variation within rural and suburban contexts that are generally automobile oriented, with few transportation alternatives available even if residents want them.
‘Modern life demands mobility, and few things are better at providing that than the automobile.’
Please lead with that next time; my brain won’t be swirling with counterpoints because I will have skipped right by.
[…] Cross-posted from Per Square Mile. […]
Interesting! I never thought about it that way.
From the paper’s, um, Abstract:
“Comparing two households that are equal in all respects except residential density, the household residing in an area that is 1000 housing units per square mile denser (roughly 50% of the sample average) will drive 1341 (6.9%) less miles per year and will consume 65 (7.0%) fewer gallons of fuel than the household in the less dense area. The joint effect of the contextual density measure (density in the context of its surrounding area) and residential density is quantitatively larger than the sole effect of residential density. A simulation moving a household from suburban to urban area reduces household annual mileage by 18%.”
Repeat: “A simulation moving a household from a suburban to urban area reduces household mileage by 18%.”
Density by itself is not what anyone serious is looking at. It must always be seen in context. Density should not be a shorthand for anyone looking at land use policy–it must include consideration of mix of uses, amenities, transportation options and other variables something like the labels “urban” and “suburban” are trying to capture
Also, isn’t 1000 units per square mile about 1.5 units/acre (with 640 acres in a square mile)? That doesnt seem like too much to get an effect
Never said density by itself was the only answer. Mixed use is definitely the way to go—and far more appealing—but I thought this study was an interesting look at how density, as it is now in the U.S., affects our driving patterns. I was surprised by the findings, but the role of employment made sense. It would be great if everyone had the option of living in a perfect, mixed-use, transit-dense Utopia (and maybe someday, most of us will be able to), but there are lots of places where that just hasn’t happened. The more I thought about it, the more I realized employment is a huge reason why so many people have to drive. Every time I have to find a new place to live, I agonize over trying to find a place where I can either walk or take transit to work. But in the future, should I (or my wife) get a new job that doesn’t jive with where we’ve chosen to live based on our current situation, we may be out of luck. I suspect that happens more often than we might think.
Also what is the impact of a person’s total transportation usage. Someone in the suburbs probably uses very little public transport. Moving to an urban environment may decrease driving but the concomitant increase in public transport may be large enough to make the total impact on the environment worse.
I don’t have a number to answer your first question, but I would suspect you’d have to use a lot of public transit to cancel out the effects of driving less. It’s just that much more efficient.
There’s a fundamental misunderstanding here (or intentional misdirection, but I kind of doubt it): anyone who works in land use understands that density – even urban density (well-discussed in the comments above) – is a *necessary but insufficient precondition* to reduced VMT.
It’s what density allows – not density itself – that reduces VMT and gets people out of cars. People would call you an idiot if you suggested that New Yorkers drive less only because New York is dense. But that’s what the study, and this article, imply.
[…] “Those who do travel by bus or train today may only be a job change away from having to drive. Modern life demands mobility, and few things are better at providing that than the automobile.” – Tim De Chant […]
[…] I’ve long had a sneaking suspicion that there must be as many cars as there are people in New York City, but it’s been a hard case to make with the popularity (and availability) of mass transit services. At Per Square Mile, Tim De Chant explains why housing density does not necessarily drive down vehicle usage. […]
[…] Pushing high density living may seem like a good way to get people out of their cars—saving them money, curbing emissions, and reducing oil dependence—but densification may not be a silver bullet, according to one recent study. The authors dug into the National Household Transportation Survey to examine per household vehicle ownership rates, vehicle miles traveled (VMT), and fuel consumption. While the results are by no means comprehensive or con … Read More […]
Almost certainly the idea that simply increasing density will reduce car usage again brings forth the mistaken idea that simple short term solutions can mitigate complex long term effects and is a result of the common planning fallacy that common location implies community – while this has been shown not to have even been true in the rural villages of past ages, we cling to these design simplicities rather than engaging with the complexity of the real city.
[…] recently cross-posted an article by Per Square Mile’s Tim De Chant which mines an old (2009) study from the Journal of Urban Economics to argue that “only the […]
It’s a bit disingenuous to say that 5000 units/mile is the density of Chicago – that may be the gross density of Chicago, but most of Chicago’s residential neighborhoods are much denser than that. I don’t think the Lakefront gets below 10000 units/mile, and that’s the exception.
I’m not paying $30 for this study, but based on the abstract, it looks like they were talking about the residential density of the neighborhoods the households they studied were in, rather than the gross density of the cities. That means their results aren’t applicable to any but the lowest density American central cities – maybe some sunbelt cities. And outside of the USA, suburbs as low density as 5000 units/mile are extremely uncommon.
Alex, while 5,000 units per square mile is the entire density of Chicago, it was the best I could come up with as a comparison and is not a terrible one, given that the study used survey data from the census block level (and many census blocks contain non-residential areas). The study focused on data from California, so no sun belt cities were included.
Speaking for myself. I live in an urban neighborhood with (according to Wikipedia) a density of 18,311/square mile. It doesn’t feel all that crowded thanks to our big hill with a public park at the top. I have a 4 mile commute to a downtown job (each way) which I complete with a bike, and get around most everywhere else by bike (faster than transit, usually). I do own a car, and mostly use it for extra-urban excursions or hauling a lot of cargo.
The net result is I drive less than 5000 miles a year, with about 25% less fuel than the national average the DOT and EPA estimate.
My neighbor who is a general contractor needs to drive his fuel-inefficient truck to various job sites, but still gains from being relatively close to them.
Another friend in a 24,740 per square mile dense neighborhood has no choice but to drive 30 miles each way daily, because the transit is too disconnected and slow.
===
What I take from this is that density works, but density and complexity work better. Density, complexity, with the right infrastructure may work best of all.
[…] approach to understanding issues of density and its impacts is compelling [see this post on density as it relates to vehicle miles traveled for a solid example], and we look forward to reading along going […]
Yup – this is terrible research. . What research shows is that – as others have pointed out, VMT goes down dramatically as you go from 1000 households per sq. mile (about 1.5 households per acre) to 10,000 households per sq mile (15 households per acre), or about the general density of Washington DC. http://switchboard.nrdc.org/blogs/kbenfield/cnt_takes_location_efficiency.html