The Intersection Between Urban Planning and Traffic Engineering Issues: Does Compact Development Truly Reduce VMT in US Cities?

While a common assertion by New Urbanists is that compact development means fewer cars on city streets, a new meta-regression analysis study has brought this association into question.

Compact, high-density development in Atlanta’s urban core has provided little to no relief from the city’s notorious issues with traffic congestion. Image Source: P. Calvert via Flickr

During the 21^st century, two of the most challenging issues faced by city and county government officials across the United States have been land use regulations and traffic congestion. Regarding zoning concerns, developers often want to maximize the density of their projects to help offset the high cost of buildable parcels in central business districts, while people who are already living in these areas usually oppose compact development, citing potential reductions in their quality of life. Caught in between these two groups are municipal elected officials, who use the guidance of the city/ town’s planning director, direct land use policy as well as approve or deny zoning variances.

When two well-respected urban planning professors, Reid Ewing and Robert Cervero, published their 2010 analysis investigating the link between land use and traffic on city streets, New Urbanists and the development community rejoiced – the researchers’ meta-analysis results indicated as compact development increased, the number of vehicle miles traveled (VMT) decreased.  Now, local elected representatives could use the promise of less traffic congestion as a carrot to coax their constituents to accept, albeit grudgingly, the approval of compact development proposals.  As time passed, some city dwellers started to question if compact development practices actually delivered on the promise of less VMT on urban streets. A new meta-analytical study, published in the January 2017 issue of the Journal of the American Planning Association, suggested a surprising answer to the public’s concerns about compact developments and city traffic.

Does Linking Between Compact Development and Vehicle Use Truly Exist?  

Mark Stevens, an urban planning associate professor at the University of British Columbia, conducted a literature review of studies addressing the link between compact development and driving behavior. He found some of the research supports the assertion that people who live in compact developments tend to drive less than those who do not, while other investigations did not find the same correlation. Given the disparities in the research findings since 2010, Stevens expressed concern that urban planners did not have conclusive evidence to promote compact development as a means to reduce driving. In an effort to reach a definitive answer to the question “Does compact development reduce vehicle use and if so, to what degree?”, Stevens examined possible reasons for the different research findings. He then conducted a meta-regression analysis of the results from 46 studies addressing the link between compact development and vehicle use.

The Issues of the Operational Definitions of Compact Development and Driving Behavior

Stevens notes that compact development has five elements:

• Density: While the public typically associates compact development with high-density projects with a large number of dwellings in a small about of space, planners use density to describe the population, the number of households, or jobs in a specified space.
• Diversity: In relationship to planning practices, the term diversity describes the variety of land uses in a particular area.
• Design: The nature and degree of delineation between corridors for walkers and drivers in the street grid of an urban area affect the way people travel and use the space.
• Destination Accessibility: The travel time between housing and jobs or residential and commercial areas is another way to assess whether planners consider that land use as compact.
• Distance to Transit: Urban planners consider residential areas within walking distance to trains, subways, or bus stations to be compact.

Since changes in any, some, or all of these aspects of compact development may affect the decisions people make about their means of travel, the differences in research findings might stem from the individual or combined effects of these 5-Ds on driving behavior. Another confounding variable is how researchers define and measure vehicle usage. For example, one study may quantify driving behavior as the number of trips a household makes per day, while another measures the amount of time individuals spend driving per week, and yet another research team uses vehicle miles travel in a discrete unit of time.

Other Factors that May Account for Different Outcomes in Compact Development Studies

In addition to the varied operational definitions of compact development and driving, Stevens cited other possible reasons for the lack of consistency in research results:

• The study had unaddressed sampling errors
• Whether or not the investigators controlled for population self-selection in their experimental design, as people may choose to live in compact developments because they prefer not to drive
• The analysis by the research team did or did not consider selective self-reporting when developing their study.

In an attempt to address the confounding variables, possible errors, as well as subject bias issues, Stevens postulated that a meta-regression analysis is a more robust statistical tool than the standard meta-analysis used in the 2010 Ewing and Cervero study. As such, he hypothesized the results from his analysis should provide clarity about the link between compact development practices and driving behavior.

Stevens Challenges the Perception that Compact Development Leads to Marked Reduction in VMT

Dense mixed use developments have limited effects on driving behavior if they are not near employers or transit options. Image Source: Brett VA via Flickr

The results from Steven’s meta-regression analysis are as follows:

• When the distance from a residence to a downtown area decreases by 1 percent, the average VMT decreases by 0.63 percent
• When the population or household density in an area increases by 40 percent, the VMT decreases by 9 percent
• When the drive time between homes and cars is between 20 to 30 minutes, the VMT is reduced by 0.20 percent.
• People tend to drive less when streets have high levels of interconnectivity.

The most surprising result of Steven’s statistical investigation is that as the mix of land uses in a discrete area increases, the average VMT also increases. This finding is in direct contradiction of the common belief in the urban planning community that mixed-use developments decrease vehicle use. Stevens also concludes that the cost of increasing urban housing density through redevelopment or infill is exceedingly high in relationship to the decrease in vehicle use. Additionally, developers and urban planners tend to experience a high degree of public resistance when infill developments and high-density housing are introduced in existing neighborhoods.

Responses to Stevens’ Study

Needless to say, Steven’s study raised a great deal of debate within the urban planning community.

One of the most notable responses to Steven’s research and conclusions came from the urban planning professors Reid Ewing and Robert Cervano, who published the 2010 meta-analysis linking compact land development with reductions in VMT. Some of the observations made in their critique, “Does Compact Development Make People Drive Less? The Answer Is Yes”, are as follows:

• Ewing and Cervano assert Stevens may have misinterpreted the results of his analysis, thus overreached in his conclusions. For example, when Stevens reports that reductions in VMT are small, they ask the question, “Small when compared to what?”
• They do not think he thoroughly examined all the benefits of compact development
• Ewing and Cervano state the sample size Stevens used in his study was too small so the results are inaccurate

Arthur Nelson, a professor of planning and real estate development at the University of Arizona also discounts Stevens’ findings in his response published in the Journal of the American Planning Association.

• Nelson observed that Stevens did not examine the effects of combining different D-variables, so his research does not investigate the cumulative effects of different elements of compact development on driving behavior.
• Nelson asserts that the high market demand for compact development means as the number of compacts developments increase, the number of VMT will potentially decrease.

Some of those who support Stevens’ findings include:

• In a 2016 interview, Gilles Duranton, a Wharton School real estate professor, reports that land use policies, such as high-density development, do little to reduce aggregate vehicle use.
• The authors of a 2017 study question if Americans are truly driving less:
  • While per capita VMT decreased in the United States dropped by 8 to 9 percent during 2004 to 2012, there is no evidence that they switched to another mode of travel.
  • During this same period, the most significant drop (12 percent) in VMT was in rural areas.
  • In urban areas, which are more likely to have compact developments than rural areas, during 2004 to 2012 VMT increased by 2.8 percent.
  • They propose that the lack of wage growth may provide a more accurate reason for the drop in VMT overall.

What is your opinion about Stevens’ research? Do you support his views or those of his critics?