Special Studies, December 7, 2009
By Paul Emrath, Ph.D and Natalia Siniavskaia, Ph.D
Report available to the public as a courtesy of HousingEconomics.com
Data from HUD and the Census Bureau show substantial differences in the way different types of households choose neighborhoods and home. The largest differences are often associated with marital status and having children. For example, married couples care more about neighborhood looks, yard, trees, view, and home size. Families with children put a greater emphasis on good schools. Single parents care most about being close to friends and family.
These often divergent location and housing preferences translate into different commuting choices. Married couples with and without children end up locating further away from jobs and having longer commutes in terms of time and miles compared to other households.
The 2000 Census data provides further insight into commuting patterns. The data show that the correlation between average block size and commute time is not linear. On average, commute time first declines and then rises as block size increases. Consequently, workers living in the most street-dense locations spend just as much or even more time commuting than those living in tracts with larger blocks.
Data on why households choose neighborhoods is available from the American Housing Survey (AHS), funded by the Department of Housing and Urban Development and conducted by the Census Bureau once every two years. This article uses data from the most recent (2007) AHS to investigate reasons households chose to live in a particular location, the basic type of housing, and several aspects of commuting behavior.
The analysis is based on households who moved recently (within the two-year period before the 2007 survey was conducted). The focus is not only on overall tendencies, but on how these tendencies differ based on household characteristics. Statistical tests often found strong and significant differences related to the type of household (married or single, with or without children, etc.). Statistical differences based on income were also frequently found; however, to make the presentation reasonably parsimonious, this article shows only AHS-based tabulations by household type. To avoid potentially misleading information, only tables are shown for which, after controlling for income, at least one household type was found to be statistically different from all others.
After controlling for household type and income, region and metropolitan status (central city, suburb, or non-metropolitan) were only occasionally found to make a statistically significant difference on the neighborhood variables under investigation. Nevertheless, due to the interest readers of Housing Economics have expressed in seeing geographic detail wherever possible, the AHS-based tables shown here include a separate set of results for each region and metropolitan category.
Neighborhood and Housing Choice
The 2007 AHS shows that the most common reasons for choosing a neighborhood are convenience to a job, being close to friends and family, and looks of the neighborhood (Table 1).
However, married couples with children under 18 value being close to schools as often as they value being close to a job. And, whether or not they have children, married couples are more often motivated by considerations that have little to do with proximity to a trip destination - considerations like the appearance of the neighborhood, or simply because they like a particular house irrespective of its location. Single parents, on the other hand, value being close to friends and family more often than any other factor, including job location (Table 2).
Table 2 shows some of the characteristics less commonly cited as a reason for choosing a neighborhood. Fewer than 8 percent of households in any category choose a neighborhood because it’s near public transportation. Married couples are less likely than other types of households to choose a neighborhood because it's close to public transportation.
Consistent with the relatively large share of married couples who choose a neighborhood because of its appearance is the relatively large share of married couples who choose a home because of its view (Table 3).
Table 3 also shows that married couples with children more often than others put an emphasis on the size of the house. In addition to their stated preferences, married couples move into homes that are on average larger, and more often single family detached - tendencies that are particularly strong if the married couples have children (Table 4).
The major differences in preferences and tendencies by household type persist across all regions and urban status categories. In contrast, there are only a few cases of urban or regional differences that persist across household types. Notable among these are the tendencies for households to move into a structure other than single family detached, or to choose a location near public transportation, if they are in the Northeast census region, or in the central city of a metropolitan area.
The AHS allows us to investigate whether reported differences in preferences for public transportation translate into actual differences in the use of public transportation or other commuting behavior.
Because married couples less often use public transportation as a criterion for choosing a neighborhood, it is not surprising that married couples end up locating further away from public transportation and use it less often as a means for getting to work. Only two percent of married couples with children use public transportation to go to work, compared to 5.2 percent of single-person and 5.4 percent of other non-traditional households (Table 5).
Since families with children care a great deal about school quality and married couples put a greater emphasis on the appearance of a neighborhood, yard, trees, view, home size, they often end up locating further away from their place of employment. This translates into the longest commutes in terms of miles and time for married couples, while one-person households have the shortest (Table 6).
Table 6 also shows that the average number of automobiles owned is highest for married couple - roughly twice that for one-person households.
For any type of household considered in this article, average distance to public transportation is shortest - and use of public transportation is most widespread - the Northeast and in central cities. Average number of automobiles owned is also lowest in the Northeast and in central cities. Otherwise, the only consistent geographic pattern across Table 6 is the short average distances to work in central cities. Irrespective of geography, married couples tend to have more automobiles and travel more miles to work than other types of households. This is generally consistent with the relatively high frequency of preferences for neighborhood characteristics that are unrelated to trip destinations expressed by married couples in Table 1.
The strong variation in locational tendencies and housing and neighborhood preferences among different types of households creates challenges for land use planning. For one thing, it suggests that the appropriate mix of housing in a transit-oriented development may be different from the mix needed to serve the local labor market as a whole.
On the other hand, use of public transportation, although relatively uncommon overall, is somewhat more common in the areas where public transportation tends to be more available - in the Northeast and in central cities of metropolitan areas. Central cities are also associated with shorter commuting distances (although not always shorter commuting times). This suggests that commuting patterns could be altered to some extent by providing public transportation or otherwise making outlying areas more closely resemble central cities.
Although it is intuitively plausible that public transportation and private vehicle use are substitutes, it does not automatically follow that building a public transportation system provides a net benefit. In addition to the commuting complexities noted above, public transportation systems such as light rail have costs - both in terms of financial resources and environmental impacts, and the costs could exceed the benefits. For example, the Portland-Milwaukie light rail is estimated to result in annual savings of 185 billion BTUs (British Thermal Units) but requires at least 2,763 billion BTUs to build it. Similarly, the environmental impact statement for Seattle’s North Link light rail estimates that the project will result in annual savings of 346 billion BTUs in 2015, declining to 200 billion BTUs in 2030 but construction will cost 17,439 billion BTUs.
Block Size and Commuting
One distinguishing trait of central cities is greater population and housing density. According to the 2000 Census, there are 2,716 people and 1,126 housing units per square mile in central cities and only 209 people and 82 housing units per square mile in parts of metropolitan areas that lie outside of central cities.
The relationship between density and vehicle use was studied in a previous article published by NAHB economists. That article found that, after controlling for other variables available in the Department of Transportation’s data set, there was a mild reduction in vehicle miles travelled as development became more compact. That article also found a "congestion" effect (vehicles being driven at less efficient speeds in more compact development), but not one large enough to totally offset the effect of reduced travel miles. So on balance gasoline consumption still tended to be somewhat lower where development was more compact.
Another feature of the built environment often assumed to have an effect on travel behavior is street accessibility and interconnectivity. The theory is that appropriately designed streets reduce vehicle commuting times and facilitate walking and bicycling. For example, a jurisdiction at the fringe of a metropolitan area that finds it undesirable, impractical, or impossible to increase overall density within its boundaries substantially, may at least be able to adopt policies designed to improve the accessibility of its streets.
The difficulty is that street accessibility and interconnectivity is not a term that is easy to define in a way that is objective or can be measured with precision. In practice, when it becomes necessary to quantify street accessibility and interconnectivity, a measure based on the size or length of blocks is almost always used. Smaller blocks suggest that streets are easier to access and better interconnected.
Block size data is relatively easy to obtain in a consistent manner across the U.S. from 2000 Census data. However, the Census does not provide micro-level information about individual households living in a block, which limits our ability to conduct an in-depth statistical analysis that controls for individual differences. On the tract level , however, the data do allow us to perform relatively simple tabulations that show the gross relationships between commuting behavior and the variable most often used as a proxy for street design. Using these data, we tabulate the average size of blocks and commuting patterns in the census tracts located in metropolitan areas .
The data show that tracts with smallest blocks (and therefore, theoretically, the best interconnected streets) have the highest percent of people walking and bicycling to work (Table 7).
Even in these street-dense tracts, however, the share of walkers and bicyclists does not exceed 10 percent. As a matter of fact, the share of bikers does not reach even one percent of commuters to work (Census data does not provide information on travel behavior except for commuting to work). As blocks get slightly larger, the share of employees walking to work drops off rapidly, and in a typical tract with an average block size of 80 thousand square meters reaches a low of 2 percent. After that, as block size increases, the share of walkers tends to stabilize and even grow a little, pointing to the share of employees who prefer and manage to locate within walking distance from work.
Similar to the percent of workers walking to work, average commute time has a U-shape response to the block size, meaning that as blocks get bigger, average commute time first declines and then rises. Workers living in the most street-dense tracts have the longest commute, approaching 30 minutes on average. To some extent this effect can be a result of a higher share of walkers and bikers in these tracts, and partially, it can be attributed to more congested streets in street-dense centers. Tracts where workers on average spend the least time commuting, less than 25 minutes, have blocks averaging 30 to 75 thousand square meters.
Data from the AHS reveal considerable differences in household preferences and behavior. Finding a location convenient to the job is an important consideration for most types of households, and this seems generally in line with land development practices that seek to locate residential development relatively close to places of employment (which does not necessarily mean the central business district of a large city).
On the other hand, households with children choose a neighborhood for schools nearly as often as for any other reason. And married couples are often interested in aspects of a neighborhood (its appearance or particular homes in it) that have little to do with proximity to trip destinations. This is consistent with the relatively long commutes and relatively low use of public transportation reported by married couples - especially married couples with children. Non-traditional (i.e., non-married couple) households more often choose a house for financial reasons.
These differences suggest that different types of housing and housing locations are needed to serve different segments of the population. The relatively low use of public transportation reported by all types of households also suggests that there are limits on how far new public transportation systems, or new homes located near current public transportation systems, can go toward reducing traffic on streets and highways. Although it is possible that increased availability of public transportation may induce more households to use it, the data suggest that the strength of this effect is likely to vary depending on the type of household.
Census data show that tracts with smaller blocks (often used as an indicator of accessible and interconnected streets) are associated with a greater percentage or residents walking or bicycling to work. However, the percentages are so small that achieving a noticeable shift away from automobile commuting through block size alone seems unlikely. As average block declines to 50,000 square feet, commuting times are declining. The trend is then reversed and commuting times are longest in tracts with an average block size less than 20,000 square feet, indicating that there are limits on how small it would be desirable to make blocks in new developments.
 Portland-Milwaukie Light Rail Project Supplemental Draft Environmental Statement, Chapter 3, p.3-155, 156.
 "Vehicle Carbon Dioxide Emissions and the Compactness of Residential Development" in Cityscape Vol 10, no. 3. 2008.
“Vehicle Carbon Dioxide Emissions and the Compactness of Residential Development” in Cityscape Vol 10, no. 3. 2008
 Use of concepts that are difficult to define and measure precisely is widespread in academic studies on land use, and often limits the number of policy implications that can be drawn from these studies.
See, for example, Ewing. R., K. Bartholomew, S. Winkelman, J. Walters, D. Chen et al. Growing Cooler. Washington, D.C.: Urban Land Institute, 2008, p.67.
Census tracts are small statistical subdivisions of a county. Census tracts usually have between 2,500 and 8,000 people. They are designed to encompass persons with similar population characteristics, economic status, and living conditions.
 We exclude very small blocks that fall below the minimum size set by the Census Bureau and those that exceed one sq. mile. Different cut-off points were tested but did not affect reported results. Blocks inside metropolitan areas are used because these are the areas that are most often the subject of land use studies. Including nonmetropolitan blocks in the analysis also did not greatly affect the results.