Research On Factors Relating To Density And Climate Change
The consensus of the scientific community is that human activity has contributed substantially to climate change through increased greenhouse gas (GHG) emissions. Given the potential impacts of a continuation of these trends, this conclusion suggests that significant action is warranted to reduce GHG emissions to avoid the worst possible consequences. One proposed course of action is to increase residential density, primarily on the grounds that it will reduce vehicle miles traveled, a measure that is closely related to the GHG emissions from driving.
Much of the vast volume of research conducted on the topic of residential density and its relationship to travel shows that there is a link between residential density and the number of vehicle miles traveled. However, the relationship is complex and characterized by inter-relationships that researchers are still in the process of disentangling. On the surface, there is a clear correlation between residential density and GHG emissions. Causation is far murkier, and this review of nearly 200 studies demonstrates that this relationship is affected by a complex set of interactions between density and at least a dozen factors, such as socioeconomic characteristics of residents, the availability of public transit, neighborhood accessibility to jobs and services, and the time and cost of various forms of transit. Although newly emerging research approaches are beginning to clarify the relationships, they are relatively untested.
This review of literature on residential density and its relationship to climate change—largely via its relationship to travel behavior—is intended to help inform this aspect of the debate on climate change by summarizing and synthesizing the literature in several key areas, discussed below.
Density is thought to influence travel behavior along several different pathways. The mode used for work travel – private auto, walking, biking, or public transit – is influenced by density at both home and work. People take non-work trips, which comprise the large majority of both trips and VMT, in order to engage in activities such as personal business, shopping, socializing, and recreation. Travel is an important part of the decision to engage in these activities, and density influences these decisions in at least three ways. Density affects the quality of the travel experience (particularly for walking), the distance required to access activities, and the price of travel, both in terms of time and money.
The research on the relationship between density and travel is virtually unanimous: after controlling for socioeconomic factors, density directly influences VMT and mode choice. However, the weight of the evidence suggests that the effect of density on travel behavior is modest (roughly 5 percent reductions in VMT and vehicle trips with a doubling of density). In comparison, large increases in regional accessibility (accessibility to regional centers), are found to have a much larger impact on travel behavior – roughly 20 percent reductions in VMT.
Based on the modest impacts on VMT of increasing density—and the difficulty of achieving that added density—several researchers suggest that it is not an effective policy tool. But some research suggests that doubling density in combination with other policies, including those that affect land-use diversity, neighborhood design, access to transit, and accessibility, could have more significant impacts on travel behavior – such as reductions in VMT on the order of 25 to 30 percent. It is important to note, however, that VMT savings will be slow to develop because of the durability of the housing stock.
Self selection is an important methodological issue that affects all studies of the relationship between travel behavior and the built environment. Researchers long assumed that characteristics of the built environment, such as density, the mix of land uses, transit availability, and neighborhood design, have a causal impact on travel behavior, the source of a significant share of the nation’s GHG emissions. More recently, researchers have re-considered the direction of causality and acknowledge that land use patterns may facilitate travel behavior but not cause it, because household decisions about residential location—and all the characteristics of this location—are simultaneous with decisions about travel behavior. That is, people who dislike driving may self-select to live in walkable neighborhoods with convenient access to transit, while people who like driving may be more likely to select neighborhoods with good auto accessibility.
An important unresolved question then is the extent to which estimates of impacts on travel behavior are affected by self selection. The weight of the evidence suggests that self selection and the built environment both have independent effects on travel behavior, but there is little research on the magnitude of the effect of each factor. Regardless, studies that ignore the impact of self selection are likely to overestimate the impact of the built environment on travel behavior. One method for correcting for self selection is to include variables in models of travel behavior that capture people’s predispositions to drive or take transit. Most studies that include these variables find that they explain a great deal of the variation in travel behavior, and suggest personality/attitudes toward driving and transit may be more important than characteristics of the built environment. However, research on this topic is in its infancy.
The size of the potential impact of changes in the built environment may depend in part on whether there is unmet demand for the high-density, walkable neighborhoods that are associated with lower auto ownership and VMT. If there is – perhaps because of local zoning restrictions that tend to encourage low density residential development – then neighborhood choices that better match consumers’ preferences could indeed result in sizeable reductions in VMT. Given this, some researchers suggest that policy makers should allow for a wide range of neighborhood types.
In addition to research on the relationship between density and travel behavior, other studies included in this review of the literature examine the influence of New Urbanism-type street patterns, demographics and income, and transit availability. The literature demonstrates that several other factors also have important impacts on travel behavior. These include trends toward business decentralization, increases in the number of two-worker households, increases in travel for non-work purposes, and increases in commercial truck traffic.
Studies that consider New Urbanism-type street patterns generally find that they have only weak or no impact on auto use. They have more impact on walking and bicycling, as does pedestrian-oriented design.
Demographic and other characteristics such as income, race/ethnicity, and immigrant status affect the degree to which residential density influences travel behavior. Other aspects of the local context – such as the local economy and geography – also affect the relationship between residential density and travel behavior. With so many factors influencing travel behavior it is clear that there is no onesize-fits-all strategy for changing travel behavior.
The general consensus of the literature is that transit availability has a negative—but marginal— impact on VMT. In general, cities with increases in transit use over the past few decades have higher population densities and are more centralized. This is consistent with findings that higher employment and population densities at trip destination increase the likelihood of using non-driving modes. Indeed, several researchers find that density at the destination is more important than at origin in predicting mode choice for work trips.
This suggests that densely developed monocentric cities with centralized employment are the best candidates for fixed rail transit. However, as discussed below, cities increasingly do not fit this description. Bus transit provides better flexibility in connecting jobs and workers than fixed-rail transit, but research consistently finds that it is more difficult to attract riders to buses than to rail transit. Given the small impacts of transit availability on travel behavior, most researchers conclude that massive investments in new rail lines would be required to substantially increase rail transit ridership and VMT.
There are at least three primary factors affecting the relationship between residential density and the climate (via travel behavior). One of these is the trend toward decentralization of employment from city centers. Less than a quarter of jobs are now located in the central business district, compared with nearly half located more than 10 miles from downtown. The trend, which started over half a century ago, indicates that the traditional view of the monocentric city is a poor approximation for the reality of most American cities. Importantly, it weakens the ability of public transit – particularly fixed rail systems – to meet travel needs, and reinforces the need for auto ownership and neighborhoods that accommodate autos.
A second factor is the increasing number of households with two workers who often commute to different locations. The literature is mixed on the implications of this trend, although there is consensus on one point: the research clearly demonstrates that households do not primarily select their residential location in order to minimize their commutes.
A third factor is the recent increase in non-work trips. Understanding trends in non-work trips is important because unlike work trips, non-work trips are often discretionary, and therefore may be more influenced by the built environment, pricing, and other factors designed to reduce auto trips and their associated greenhouse gas emissions. On the other hand, non-work trips may be less influenced by public transit options because they often involve multiple destinations and are thus less well suited to public transit than work trips.
Trends over the past decade also indicate that commercial truck traffic is increasing its share of total VMT, and that this trend is likely to continue in the next decade.
In addition to unresolved questions about the role of self selection, other important questions are left unanswered by the current research on the connection between residential density and the climate. Among others, how difficult would it be to achieve residential densities that are double their current levels across a metro area – that make Atlanta look more like Boston? Experience from Portland, Oregon, an area known for its urban growth boundary, suggests that sizeable increases in density takes decades – at least 30 years. Given that the built environment is long lived, this result is not surprising.
Few studies include the impact of travel cost—either in terms of time or money—on travel behavior, but those that do conclude that pricing may play a more important role in explaining travel behavior than characteristics of the built environment. They conclude that changes in policies that affect the monetary or time cost of car ownership and use—such as increases in gas taxes or the price or availability of parking and the supply of roads—are more effective in changing travel behavior than any other policy. If policy makers find these types of economic incentives to be unpalatable, policies that lead to large-scale changes in land-use are a distant second-best alternative.