The Role of Smart Growth in Residential Location Choice: Heterogeneity of Location Preferences in the Chicago Region

Abstract

Spanish

Chinese

We investigate the role of smart growth in household location choice in the Chicago region, using a discrete choice analysis. In the midst of continued region-wide suburbanization, households tend to move to neighborhoods with rich consumption amenities and high transit access. However, this study does not find evidence that the neighborhood’s physical compactness is a significant location factor. Location preference for compact, mixed-use, and transit-oriented neighborhoods is significantly affected by the life cycle stage and income level, but less influenced by generation and age. Millennial households show strong preference for amenities and transit access only before they marry and have children.

Keywords

location choice consumption amenities millennials smart growth urban form

Introduction

Since the mid-twentieth century, low-density single-family home development has dominated the landscape in U.S. cities and suburbs, leading to many urban problems including high car-dependency, excessive infrastructure costs, traffic congestion, pollution and greenhouse gas (GHG) emissions, and social segregation (Burchell et al. 1998). Major planning approaches including new urbanism, smart growth, and transit-oriented development have emerged in part to engage with these issues. These initiatives all aim to reconfigure urban neighborhoods into more compact, mixed land use, pedestrian- and transit-friendly urban form (Fulton 1996; Knaap and Talen 2005; Song and Knaap 2004). Previous studies refer to these strategies as forms of “alternative devleopment” or “alternative neighborhoods” for such urban form (Cao 2008; Levine and Inam 2004). Today, most recent development projects across the country have more or less incorporated some elements of smart growth into their plans (Kalinoski 2018). Given the shared aims of planning strategies, we use the terms smart growth and compact, mixed-use, and transit-oriented development interchangeably throughout this article.

Advocates for smart growth argue that there is a strong unmet demand for compact, mixed-use, and transit-oriented neighborhoods that are undersupplied largely due to municipal regulations such as exclusionary zoning and auto-oriented urban design standards (Levine 2005; Levine and Inam 2004). If this is the case, compact, mixed use, and transit-oriented development would alleviate planners’ burden to prove travel and environmental benefits of such development (Levine, Inam, and Torng 2005). Regardless of the extent to which compact development promotes sustainable travel, smart growth should be promoted in its own right to offer enough residential options to American households with diverse location preferences. Yet, this “choice-based rationale” for smart growth has not been fully validated because existing studies offer mixed evidence as reviewed in the next section (Cao 2008; Handy et al. 2008).

Research to date has mainly relied on household surveys (Cao 2008; Handy et al. 2008; Levine and Inam 2004), demographic forecasting (Myers and Gearin 2001), and hedonic modeling of home prices (Bartholomew and Ewing 2011; Song and Knaap 2003) to investigate American household’s current and future preferences for compact, mixed-use, and transit-oriented neighborhoods rather than examining actual location choices. However, stated preference in surveys often fails to capture actual behavior, and hedonic pricing studies cannot take into account the heterogeneity of preferences among different types of households. Discrete choice modeling is well-suited for studying the discrete nature of location choices when appropriate data at an individual household level are available (Chen, Chen, and Timmermans 2008; Schirmer, Van Eggermond, and Axhausen 2014). However, existing discrete choice studies often fail to consider two important aspects of location preference: trade-off between the built environment and other traditional location factors such as job accessibility and housing price; and the heterogeneity of households’ preferences for the built environment. One exception is a relatively rich body of literature that specifically focuses on low-income households and subsidized housing, for whom the mobility data are more readily available (Basolo and Yerena 2017; Kleit and Galvez 2011; Tremoulet, Dann, and Adkins 2016). Studying heterogeneous preferences for compact, mixed-use, and transit-oriented neighborhoods is particularly important in light of the emerging location choices of Millennials or today’s young adults toward central cities and urban neighborhoods (H. Lee 2020; Moos 2014).

To address these gaps, we investigate the role of compact built environment and urban amenities in residential location choice in the Chicago urbanized area (UA), using a series of multinomial logit (MNL) regression models and unconventional data sources. In doing so, we also examine heterogeneous preferences for compact, mixed-use, and transit-oriented neighborhoods across different types of households by life cycle stage, age, generation, and income. Household residential location data come from a unique commercial dataset—InfoUSA U.S. Consumer Database. This database consolidates and imputes information on a range of individual and household characteristics, including residential addresses. It is longitudinal in nature and thus we can use these data to identify household residential location changes over time. It covers a considerably large sample of individual households in the study region and is much less likely to suffer from selection bias.¹ Moreover, we operationalize neighborhood-level smart growth attributes in two dimensions, physical characteristics of the built environment and urban consumption amenities. We use some of the key variables that are developed applying a web scraping technique to Google Maps Application Programming Interface (API; Y. Lee, Lee, and Shubho 2019).

Our research confirms that the built environment is complex and multi-dimensional. Different elements of smart growth play different roles in household location choice. Households tend to decentralize in general, but they still want urban characteristics in their neighborhoods, such as the presence of recreational opportunities and public transit. However, our model shows that physical compactness is not significantly associated with households’ location outcomes in the study area. The results also confirm that location choice for compact, mixed-use, and transit-oriented neighborhoods is largely influenced by the household’s life stage and income level, but less by age and generation. Those who value these smart growth elements are more likely to be single households without kids. When households step into later life cycle stages, the attractiveness of smart growth is greatly reduced. Low-income households are less likely to relocate into neighborhoods with greater urban amenities, perhaps due to their lower affordability.

Literature: Smart Growth and Residential Location Choice

Market for Compact, Mixed-Use, and Transit-Oriented Neighborhoods

While low-density and single-use development in urban landscapes reigned for decades, several progressive planning initiatives, such as new urbanism, smart growth, and transit oriented development (TOD) have emerged in recent years, proposing alternative built environment designs. While a great diversity of design practices have been proposed at different spatial scales, most of these emerging development principles share common features: compactness and mixed land use, mixed-income development, pedestrian-friendly streets, accessible transit, and well-designed public spaces (Ellis 2002; Trudeau 2013). These components, highlighted in the Charter of the New Urbanism published by Congress for the New Urbanism (CNU), are believed to promote sustainable and healthy travel behavior and urban life. However, many scholars pointed out that while a neo-traditional aesthetic style is widely adopted in New Urbanist projects, “true” smart growth design practice (including streetscapes, public spaces, and densities) and land use policies (mixed-use, mixed-income, mixed-tenure, and TOD) are much less commonly adopted (Ellis 2002; Sohmer and Lang 2000; Talen 2000; Trudeau 2013; Trudeau and Malloy 2011).

Researchers’ assessments of smart growth are mixed (Ellis 2002; Talen 2000). Urban economists have questioned the ideological land use principles of new urbanism as they debate what characterizes a good urban form and if accessibility in fact matters in city life (Gordon and Richardson 1997, 1998, 2001). In their view, low-density suburban development reveals consumers’ preferences in a free market and the demand for compact, mixed-use, and transit-oriented neighborhoods is minimal (Gordon and Richardson 1997, 1998). They reject the idea of planning intervention that imposes alternative development on urban space where the market is working in the opposite direction. Other planning scholars argue that existing development patterns are not just an outcome of market choices, but also the result of many regulation barriers to smart growth development, including exclusive zoning and subdivision regulations, auto-oriented street design, and excessive parking standards (Levine 2005; Levine, Inam, and Torng 2005). Indeed, a nationwide survey of developers reports that inadequate supply of smart growth development is mainly due to local government regulations and substantial proportions of their proposals for such development were either rejected or significantly altered by local governments (Levine and Inam 2004).

Is there a market for smart growth and is it growing? To address this important question, most existing studies have analyzed household preferences that are stated in housing surveys or revealed by home prices. Many studies based on household or consumer surveys show that there is a substantial size of unmet demand for (neo-) traditional or pedestrian-/transit-oriented neighborhoods (Handy et al. 2008; Levine and Inam 2004; National Association of Realtors 2015; Urban Land Institute 2013); while some others show that transportation and land use characteristics are secondary to housing price, neighborhood safety, and many other place-related attributes in household location decisions (Cao 2008). Furthermore, Myers and Gearin (2001), combining survey evidence and demographic projections, predict that the demand for smart growth will continue to grow in coming decades.

Another group of studies have investigated housing prices to see if the features of compact, mixed-use, and transit-oriented development are capitalized into home prices, revealing market preferences. Most of these studies use hedonic regression models to tease out the monetary values of neo-traditional neighborhood attributes. For example, an analysis of housing prices in Washington County, Oregon, found that residents are willing to pay about a 15 percent premium to buy a home in a neighborhood with new urbanist features such as connected street, walkability, and proximity to light rail stations (Song and Knaap 2003). But, the same study also found that density and land use mix have negative impacts on single-family home prices. A meta-analysis of hedonic analyses shows that the impacts of local proximity to rail stations on property values are generally larger for commercial properties than residential properties (Debrezion, Pels, and Rietveld 2007). Another more comprehensive review of the literature suggests that the amenity-based elements of pedestrian- and transit-oriented development, beyond accessibility benefits, are being capitalized into urban land and home prices (Bartholomew and Ewing 2011).

These studies using housing surveys and home prices have a common limitation in that they do not examine actual location choice behavior. In particular, housing survey studies typically examine the stated preferences for hypothetically set built environments that are often bundled with housing characteristics and often include unrealistic scenarios (Cao 2008; Myers and Gearin 2001). Furthermore, surveys of stated preference tend to overstate the value of or a willingness to pay (WTP) for public goods due to many reasons, including social desirability and affirmation biases—respondents may give answers that are socially desirable or the interviewer wants to hear (Loomis 2014). This is one of the reasons why consumers who consider owning a car, privacy from neighbors, and light traffic important location factors still show strong and growing support for (neo-) traditionally designed communities in surveys (Handy et al. 2008).

A critical limitation inherent to hedonic pricing studies is that they fail to identify the heterogeneity of preferences among different types of households. What types of households are more likely to choose compact, mixed-use, and transit-oriented over suburban single-family neighborhoods? The hedonic price models cannot address this important question in light of demographic shifts, including retiring baby boomers and surging Millennials. To overcome these limitations of existing literature, the current study attempts to directly investigate actual location decisions of households.

Residential Location Choice and Compact, Mixed-Use, and Transit-Oriented Development

Residential location decision of a household determines its accessibility to the rest of the city and greatly affects the way household members travel around the city (Guo and Bhat 2007). At the aggregate level, residential location choice drives travel demand and influences business location decisions, spatial segregation, and the urban transportation system (Schirmer, Van Eggermond, and Axhausen 2014). Therefore, understanding the connection between the built environment and household location choice is very important in making better housing, land use, and transportation policies.

However, the built environment received little attention in early location studies. Urban economic models developed by Alonso (1964) and others (Mills 1967; Muth 1975) offered an influential framework to understand a household’s location behavior within a city, in which a utility maximizing household chooses its location by weighing commuting costs against housing costs. While the economic models highlight the utility maximization principle and the importance of job accessibility in residential location choice, they have been criticized for the assumption of homogeneous households and urban space (Montgomery and Curtis 2006).

Discrete choice models pioneered by McFadden (1978) provide a powerful framework in which researchers can consider a substantially expanded list of location factors. Now, beyond a proximity to the urban center, location in urban space can represent the built environment, socioeconomic environment, points of interest, and accessibility (Schirmer, Van Eggermond, and Axhausen 2014). A household trades off job access with many important neighborhood attributes besides housing costs (Hu and Wang 2019; T.-K. Kim and Horner 2003). Furthermore, one can take into account the characteristics of the household who makes the location decision and can explore the different location preferences among different household types by size, income, race/ethnicity, life cycle stage, and perhaps generation.

One important location attribute in the intra-metropolitan context is public services. According to the “Tiebout hypothesis” (Tiebout 1956), households “vote with their feet” based on the diverse packages of municipal tax and services that different cities offer. While the evidence on the impacts of general public goods on household location choice is mixed (Friedman 1981; Reschovsky 1979), the quality of public schools has been consistently found to be a significant location factor particularly for households with school-age children (Bayoh, Irwin, and Haab 2006; J. H. Kim, Pagliara, and Preston 2005; Zhou and Kockelman 2008). It should also be noted that the fragmented local government system in the Tiebout model has facilitated racial sorting and segregation in U.S. regions (Dawkins 2005; Howell-Moroney 2008).

Demographic and socioeconomic characteristics of the neighborhood are also very important location factors as households tend to choose a location where a similar group of households are around. These neighborhood attributes include racial/ethnic composition (Gabriel and Rosenthal 1989; Gottlieb and Lentnek 2001; Pinjari et al. 2011), household age, size and types (Guo and Bhat 2007; B. H. Y. Lee et al. 2010; Pinjari et al. 2011), and household income (de Palma, Picard, and Waddell 2007; Weisbrod, Lerman, and Ben-Akiva 1980). Lindstrom (1997) suggests that housing and location choice can be viewed as a way to express their social status and identity by co-locating with people who share the same values and preferences.

Households also consider natural amenities such as the access to scenic views, open space (Hörnsten and Fredman 2000; T.-K. Kim, Horner, and Marans 2005), and recreational sites (Colwell, Dehring, and Turnbull 2002). The desire to live close to the nature and open space is often cited as one of the reasons behind the sprawling development in urban fringe (Kaplan and Austin 2004). With the emergence of the Millennial generation, however, the role of urban, cultural, and consumption amenities are increasingly important in location choice studies. While earlier studies used the distance to the urban center as a very crude proxy for urban neighborhood characteristics (Andrew and Meen 2006; Axhausen et al. 2004; de Palma, Picard, and Waddell 2007), more recent efforts directly seek to measure consumption amenities and new urbanist neighborhood attributes (Couture and Handbury 2017; Y. Lee, Lee, and Shubho 2019).

The role of the built environment in residential location choice is generally understudied. Nonetheless, one of the most common findings in the literature is that there is great heterogeneity in the preferences for certain neighborhood attributes, including the features of smart growth development. For example, high density, a hallmark of smart growth, is generally not favored by high-income households and large families; but young, employed single households prefer high density and city living (de Palma et al. 2005; Guo and Bhat 2007; B. H. Y. Lee et al. 2010; Pinjari et al. 2011; Zondag and Pieters 2005). Land use mix also appeals to some types of households, such as young households without a car, more than others (Guo and Bhat 2007). Several recent studies show that central locations with rich urban amenities has emerged as a competitive option for senior households at later stages of the life cycle as well (Marois, Lord, and Negron-Poblete 2018; Walker 2016b).

Last, an increasing number of studies pay attention to a possibility of generational shifts in residential location choice. Millennials, born between 1985 and 2000, are known to buy fewer cars, make fewer trips, and use alternative modes of travel more often than previous generations (Blumenberg et al. 2016; Davis, Dutzik, and Baxandall 2012; Manville, King, and Smart 2017; McDonald 2015; Zhong and Lee 2017). Paralleling this trend, urban lifestyles of a significant segment of Millennials, largely influenced by economic restructuring and societal changes, are driving the resurgence of high-density, central city neighborhoods with better transit access and walkable streets (Moos 2014). Several studies note that consumption and cultural amenities, more than the physical environment in urban neighborhoods, are driving the urban resurgence by young and educated population in recent years (Couture and Handbury 2017; Y. Lee, Lee, and Shubho 2019).

In sum, recent studies using housing preference surveys and hedonic analysis of home prices indicate that there is a growing market for compact, mixed-use, and transit-oriented development, but these studies do not analyze actual location choice behavior. While discrete choice models offer a powerful framework to study a long list of location factors beyond simple job accessibility, the location choice literature has not fully engaged in studying the role of the built environment and urban amenities in residential location choice largely due to data constraints.

Method

Study Area and Data

We selected the Chicago UA as the study area. While the census-defined UA boundary spans from Illinois to Northwest Indiana with approximately 8.6 million residents in 2010, we only include census tracts within the State of Illinois. Nearly 30 percent of the population in the region live in the city of Chicago, which is the third most populous city in the United States. The Chicago region includes diverse communities and neighborhoods, including financial and cultural center in downtown, working-class and relatively affluent neighborhoods in the north, struggling neighborhoods in the Chicago south, and many older and newer suburbs in the outer-ring of the region. Chicago is also culturally diverse with many ethnic enclaves scattered throughout the region. The entire region has been rapidly changing with many neighborhoods being gentrified, diversified, and transformed (Greenlee 2019). Such diverse neighborhoods present a great choice set for individual households to choose residential location from, thus is well-suited for studying household location choice.

We used a unique dataset for studying household location choice that is rarely used in the literature: InfoUSA U.S. Consumer database. This dataset, developed for commercial use,² records longitudinal geographical location and demographic information of 155 million households nationwide. The data are assembled and updated monthly by drawing residential location information from United States Postal Service National Change of Address database, Locatable Address Coding database, and Delivery Point Verification database. For the demographic and socioeconomic information, InfoUSA draws upon other sources of public records, for example, deed transfer and tax assessor information. InfoUSA uses a proprietary methodology to derive individual and household characteristics from these data sources. Researchers have estimated the gross national undercoverage rate for InfoUSA data at 8.4 percent, with an undercoverage rate of 18.8 percent for Illinois households (Kennel and Li 2009). This level of undercoverage is comparable with that assessed by other private data sources, where gross national undercoverage rates range between 15 and 23 percent (Dohrmann, Han, and Mohadjer 2006; Iannacchione et al. 2007; O’Muircheartaigh, Eckman, and Weiss 2002; O’Muircheartaigh et al. 2006).

We defined a residential move as the change of the longitude and latitude of residential address between two subsequent years. We limited the data sample only to intraregional moves, where both the origin and the destination are located within the study area. Using InfoUSA, we derived a dataset featuring 886,986 residential moves within the study area from 2006 to 2015, which offers a much richer set of household mobility information compared with conventional data sources. While we used descriptive mapping to visualize all the 886,986 residential moves, MNL regression models used randomly sampled 8,000 moves, 2,000 moves for each of four life cycle groups (single households without kids, single households with kids, married households without kids, and married households with kids), for a computation limit. Although the InfoUSA consumer dataset contains longitudinal records, the amount of time recorded for individual households varies. Therefore, the dataset is treated as a cross-sectional data for this research, with each residential move being a single observation.

Discrete Choice Modeling

We investigate residential location choice of diverse households in the Chicago region, using a MNL model which is popular in transportation and housing studies (Ben-Akiva and Lerman 1985; McFadden 1978). In this discrete choice model, a household $n$ in the sample $N$ is assumed to make a decision within a choice set $L$ that includes k number of neighborhood choices such that the chosen neighborhood i offers greater utility than alternative locations $l_{1}, l_{2}, \dots, l_{k - 1}$ (i.e., utility maximization):

U_{i n} > U_{j n}, for all j \in L and i \neq j .

U_{i n} = V_{i n} + ε_{i n}, where V_{i n} = α X_{i} + β H_{n} X_{i}^{'} .

The utility $U_{i n}$ that a household $n$ gains from a neighborhood i with certain characteristics is assumed to consist of a deterministic component $V_{i n}$ and a random part $ε_{i n}$ that follows a certain probability distribution. The deterministic term of the utility is assumed to be a linear function of neighborhood characteristics $X_{i},$ including accessibility, physical urban form, socioeconomic attributes, and urban amenities. The MNL model assumes an independent and identically distributed (IID) Gumbel distribution, and the probability for the household to choose neighborhood i is

\Pr (i) = \frac{e^{V_{i n}}}{\sum_{j = 1}^{L} e^{V_{j n}}} .

The log-likelihood function is thus given as $LL = \sum_{n = 1}^{N} \sum_{i = 1}^{L} d_{i n} \ln (\Pr (i)), d_{i n} = 1 if i being chosen, = 0 if not .$

The coefficient vectors $α$ and $β$ are estimated by maximizing this log-likelihood function.

One critical choice to make in MNL modeling of location decision is the creation of alternative choice sets. It is unrealistic and problematic to generate a choice set that includes all census tracts in each UA because it not only adds irrelevant choices to households but also violates the IID assumption of the error term (Cho, Rodriguez, and Song 2008). To create a proper choice set, some researchers have adopted random sampling from all possible alternatives in the area (Guo and Bhat 2001), while others have applied deterministic approaches, considering distance to employment or housing affordability, for more consistent and presumably unbiased estimations (Cho, Rodriguez, and Song 2008). However, InfoUSA data do not provide workplace location information. It is also challenging to determine housing affordability for households that are temporarily unemployed or out of labor force. Balancing the efficiency and accuracy of the model, and accounting for the unique structure of the InfoUSA data, we give each household a choice set of five neighborhoods (census tracts), including the actual choice and four alternative locations that are randomly selected within a ten-mile radius of the actual choice.

We first run a set of base models that investigate the role of compact, mixed-use, and transit-oriented neighborhood attributes in location outcomes of households at four different life cycle stages. To test how preferences vary across different population groups, we interact our key neighborhood-specific variables with household characteristics, including generation, age, and income. For the generation effects, we define Millennials, Generation Xers, and Baby Boomers as those who were born between 1985 and 2000, 1965 and 1984, and 1945 and 1964, respectively,³ and use Generation Xers as a reference group. For the age effects, we divide the entire sample into three age groups, less than thirty years old, thirty to fifty years old, and greater than fifty years old. And last, for the income effects, we define low-income and high-income households as those whose income falls into the bottom 25 percentile and top 25 percentile, respectively, and use the rest middle-income households as a reference group.

Measuring Smart Growth Attributes

We operationalize smart growth attributes or neighborhood-level urbanism in two dimensions: physical characteristics of the built environment and consumption amenities. Extensive efforts in recent years to define the physical components of compact development that facilitate sustainable urbanism converge on the “Five Ds”: density, diversity, design, distance to public transit, and destination access (Ewing and Cervero 2010). We use census tract–level compactness index developed by Ewing and Hamidi (2014) and two additional variables to quantify neighborhood-level physical urban form. Ewing and Hamidi derived the compactness index from multiple variables capturing density, land use mix, walkability, and street connectivity using principal components analysis. We add two additional variables as the compactness index captures only neighborhood-level urban form and land use (the first three Ds), but not location information in the regional context and transit access (remaining two Ds). We add a transit-based job accessibility index and the distance from Chicago downtown. The transit-based job accessibility index is developed using the framework suggested by Shen (1998) and the General Transit Feed Specification (GTFS) data from major transit agencies in the Chicago region. In particular, this index takes into account the number of feasible alternative transit routes and their frequency in addition to travel times (see J. Kim and Lee 2019, for the method in detail). We expect this variable to capture both transit availability and access to employment opportunities. The distance to Chicago downtown is included to detect overall decentralization or re-centralization trend in the region in recent years.

Consumption amenities in urban neighborhoods are also an important component of smart growth. In particular, studies show that recent generations such as late Generation Xers and Millennials value various shopping, entertainment, recreational, and social opportunities in their location choice (Couture and Handbury 2017; Y. Lee, Lee, and Shubho 2019). Glaeser et al. (2001) define consumption amenities as a variety of specialized goods and services, aesthetic and attractive urban setting, and high-quality public services that are typically found in large urban centers. Researchers have just begun to operationalize consumption amenities. Couture and Handbury (2017) developed a consumption amenities index based on the number of retail, entertainment, sports, restaurant, and other service establishments. Following this approach, we develop a census tract–level index for consumption amenities based on the density of commercial, social, and cultural places, including cafes, bars, nightclubs, museums, art galleries, and shoe stores. Google Maps API provides business information for nearly one hundred types of businesses such as business type, hours, quality rating, and location. We measure the density of these business in each census tract and then develop a composite index of overall urban consumption amenities.

One caveat for applying this method is that the data were scraped from the Google Maps API in 2017, while the moves identified in the InfoUSA data set occurred between 2006 and 2015. This inconsistency raises the issue of endogeneity: neighborhood characteristics may change because of the influx or outflux of households, even though the changes in local amenities tend to be slow. To assess the impacts of this potential issue, we ran exactly the same models with only three recent year (2013–2015) data sample. The results were largely consistent with those of ten-year data models except that the physical compactness variable became significant at 5 percent with a very small negative coefficient for married households. We decided to present ten-year data models in this paper because we need to use the data for a long-enough period to separate out generation effects from age effects. In any case, we treat the result for this variable as a depiction of correlation rather than causality between location choice and consumption amenities.

Figure 1 presents the distribution of urbanism variables in the study area. The compactness index shows a combination of concentric ring and radial pattern in which values decrease with the distance from Chicago downtown, but showing some moderate to high levels along major transportation corridors. The consumption amenities index also has its highest values in central Chicago, but is more evenly distributed throughout the region, with some significant hot spots outside the central city. Many suburban downtowns and edge cities appear to maintain a decent level of consumption amenities. We view physical compactness of the built environment and consumption amenities as complementary and both are important components of smart growth development. This study examines how each component influences household location outcomes.

Figure 1.

Spatial patterns of neighborhood urbanism in the Chicago urbanized area: (A) Compactness Development Index and (B) Consumption Amenity Index.

Besides these key urbanism variables, we control for other location factors, including housing market supply, neighborhood socioeconomic status, and school quality at the census tract level. We additionally include percent married family and percent households with children to control for the tendency for households to move to neighborhoods with people at a similar life cycle stage. As some of the neighborhood attributes may appeal to different household types in different ways, we interact census tract–level median income and percent owner-occupied units with corresponding household-level variables, respectively. We build a census tract–level school quality measure based on school ratings scraped from www.greatschools.org that derive school ratings from diverse inputs such as test scores, student progress, and equity rating. It captures the quality of elementary schools in each census tract as of January 2017 (Y. Lee, Lee, and Shubho 2019). Table 1 describes all the variables used in the analysis and data sources.

Table 1.

Variable Description and Data Sources.

Variable name	Description	Data source
Alternative development variables
Distance to downtown	Natural logarithm of physical distance to Chicago downtown	TIGER GIS Data
Compactness index	Composite index for physical compactness: Principal components extracted from multiple variables capturing density, land use mix, walkability, and street connectivity	Ewing and Hamidi (2014)
Amenities index	Composite index based on the density of places for commercial, social, and cultural activities—Cafes, bars, nightclubs, museums, art galleries, and shoe stores	Lee et al. (2019)
Job accessibility by transit	Gravity-type transit-based job accessibility index constructed using Shen’s (1998) framework and GTFS data. The index uses the number of feasible alternative transit routes and their frequency in addition to travel time as travel impedance	J. Kim and Lee (2019)
Control variables
% married family	Married families’ percent share of all the households within the census tract	ACS five-year estimates
% HH with children under eighteen	Percent share of all households that have children aged under eighteen within the census tract	ACS five-year estimates
Elementary school quality	Average of elementary school ratings scraped from www.greatschools.org that derive school ratings from diverse inputs such as test scores, student progress, and equity rating	Lee et al. (2019)
Job accessibility by auto	Automobile-based job accessibility constructed using the same framework as the one for transit-based accessibility	J. Kim and Lee (2019)
% multi-family units	Multi-family units’ percent share of all the housing units within the census tract	ACS five-year estimates
% owner-occupied units	Owner-occupied units’ percent share of all the housing units within the census tract	ACS five-year estimates
Log (median income)	Natural logarithm of median household income for households within the census tract	ACS five-year estimates

Note: GTFS = General Transit Feed Specification; ACS = American Community Survey; HH = households.

Spatial Pattern of Residential Moves in the Chicago Region

The kernel density maps in Figures 2 and 3 present the spatial distribution of over 800,000 residential moves between 2006 and 2015 in the Chicago UA. Figure 2 shows all the moves in one map while Figure 3 presents residential location choices by household type. Overall, the geography of intraregional moves resembles the spatial distribution of current population in the region. The densities of both origins and destinations of moves are the highest in downtown Chicago and show some radial pattern along major transportation corridors. There are also several hot spots around major suburban centers such as Schaumberg, Naperville, and Aurora.

Figure 2.

Kernel density maps of origins and destinations of residential moves in the Chicago UA: (A) Destinations and (B) Origins.

Figure 3.

Destinations of residential moves by household type.

However, as shown in Figure 3, great heterogeneity in residential location choice emerges as we break down the sample into household groups of different income status and life cycle stages. First, the moves of low-income households, regardless of their stages in life cycle, are geographically confined to a small portion of the study area, central and south Chicago; while households with medium to high income enjoy expanded ranges of housing and neighborhood options, showing much dispersed moving patterns. Except for single households without kids, medium- and high-income households’ residential locations nearly cover the entire study area except that high-income households tend to avoid low-income neighborhoods in the south.

Second, the stage in life cycle also greatly influences household location choice. Single households without kids, who tend to be young, stand out in their residential location choice. In both middle- and-high-income groups, their moves were extremely concentrated in downtown Chicago and the vicinity, where employment opportunities are the highest and diverse urban amenities are rich. Either marriage or the presence of kids dramatically expands their choices toward suburban locations where larger home space is more affordable. In the next section, we analyze how these household profiles, income and life cycle stage, affect their location preferences for compact, mixed-use, and transit-oriented neighborhoods and consumption amenities.

Location Choice Modeling Results

Preference for Smart Growth and Life Cycle

Our results confirm the complexity of household preferences for compact, mixed-use, and transit-oriented neighborhoods. Table 2 shows MNL model results by household type in terms of life cycle stages. Estimated coefficients reveal how neighborhood attributes affect a household’s decision to move to the chosen neighborhood as opposed to four randomly selected alternative locations.

Table 2.

Multinomial Logit Analysis Results for Base Models by Life Cycle Stage.

	Model 1SN	Model 2SK	Model 3MN	Model 4MK
Built environment variables
Log (distance to downtown)	0.56***	1.05***	1.12***	1.06***
Log (distance to downtown)	(0.09)	(0.11)	(0.12)	(0.12)
Compactness index	–0.003	0.001	0.001	–0.004
Compactness index	(0.002)	(0.002)	(0.002)	(0.003)
Amenities index	0.20***	0.16***	0.15***	0.10**
Amenities index	(0.04)	(0.04)	(0.04)	(0.04)
Job accessibility by transit	0.30***	0.28***	0.16**	0.14*
Job accessibility by transit	(0.07)	(0.07)	(0.08)	(0.09)
Control variables
% married family	–0.01**	–0.01***	0.02***	–0.003
% married family	(0.004)	(0.003)	(0.004)	(0.004)
% HH with children under eighteen	–0.03***	–0.01***	–0.01***	–0.004
% HH with children under eighteen	(0.004)	(0.003)	(0.004)	(0.004)
Elementary school quality	0.02*	–0.03**	–0.02	–0.01
Elementary school quality	(0.01)	(0.01)	(0.01)	(0.01)
Job accessibility by auto	0.06	–0.04	–0.02	–0.12*
Job accessibility by auto	(0.06)	(0.06)	(0.06)	(0.07)
% multi-family units	0.63***	0.38**	1.00***	0.45**
% multi-family units	(0.18)	(0.17)	(0.20)	(0.20)
% owner-occupied units × HH home-owners	2.47***	3.43***	2.48***	2.55***
% owner-occupied units × HH home-owners	(0.40)	(0.35)	(0.30)	(0.30)
Log (median income) × High-income HH	2.84***	2.76***	2.05***	1.85***
Log (median income) × High-income HH	(0.30)	(0.32)	(0.16)	(0.15)
Log (median income) × Low-income HH	–0.67***	–1.06***	–1.86***	–2.03***
Log (median income) × Low-income HH	(0.11)	(0.12)	(0.22)	(0.31)
Observations	2,000	2,000	2,000	2,000
Log likelihood	2,901.09	2,944.30	2,846.88	2,837.96
AIC	5,826.19	5,912.61	5,717.75	5,699.92

Note: SN = single and no kids; SK = single and with kids; MN = married and no kids; MK = married and with kids; HH = households; AIC = Akaike information criterion.

p < .1. **p < .05. ***p < .01.

Overall, households in the Chicago region are still likely to move to locations farther away from downtown, all else being equal, which is consistent with aggregate decentralization trend in the region. As expected, this tendency of relocation toward more decentralized locations is the strongest among married households with kids that generally need larger home space and is the weakest in the model for single households with no kids.

The preference for other smart growth elements shows mixed results. Households of all types value rich consumption amenities and job accessibility by transit while their residential location choice is not significantly affected by physical characteristics of the development. The compactness index, mainly capturing density and street network connectivity such as block size and intersection design, turns out to be insignificant in attracting all four types of households, regardless of marriage status and the presence of children.⁴ Instead, the density of commercial, social, and cultural places and availability of public transit are significant location factors in all four models. As job accessibility by automobile shows no significance (or a negative sign among married households with kids), we interpret the job accessibility by transit variable is capturing transit access more than job density.

The results for control variables show that, as expected, households are likely to choose neighborhoods where households of similar status are concentrated. Single households are likely to choose neighborhoods with lower shares of married couples, and households with no kids tend to avoid places where households with kids are high. Potential home-owners look for places with more home-buying options, and households tend to move into neighborhoods whose residents are of similar income levels. The availability of multi-family housing in the neighborhoods attract all households and the effects are much stronger for households without kids that generally have a lower demand for home space.

The only variable showing an unexpected result is elementary school quality, which has a positive sign in the model for single households without kids, a negative sign for single households with kids, and has no significance for married households. We assume it’s because households in the study area have options to go to schools across neighborhoods, and thus the local school quality becomes less relevant.

A stage in life cycle strongly affects the way households trade off smart growth components and other location factors. Table 2 shows that single households without kids (model 1) are most likely to favor urban neighborhoods, with the smallest coefficient for distance to downtown (+0.56), and the largest coefficients for consumption amenities (+0.20) and transit-based job accessibility (+0.30). As households step into later life cycle stages, their demand for space grows and their perception and preference for various neighborhood attributes change. Both coefficient size and significance level in Table 2 are consistent with these changes across life cycle stages, with married households with kids showing the weakest preference for consumption amenities and transit access (model 4).

To further statistically test the differences in coefficient size across the household types, we reran a MNL model with a pooled dataset that includes interaction terms of life cycle stage dummies and the four smart growth variables. Results, presented in Table 3, are consistent with those in Table 2. Compared with single family with no kids (used for a reference group), getting married significantly reduces the desirability of transit-based job accessibility because married couples often need to compromise their travel needs. The presence of children in the family significantly weakens the preference for consumption amenities, as the lifestyle of families with kids tend to prioritize other neighborhood attributes such as safety, quietness, and availability of open space and neighborhood parks.

Table 3.

Statistical Test of the Difference in Coefficient Sizes Across Life Cycle Stages.

	Base model with pooled data
	SN	SK	MN	MK
Log (distance to downtown)	(reference)	0.923***	0.280	0.464
Log (distance to downtown)		(0.267)	(0.271)	(0.292)
Compactness index		0.008	–0.008	–0.007
Compactness index		(0.007)	(0.006)	(0.007)
Amenities index		–0.27***	–0.156	–0.199*
Amenities index		(0.103)	(0.104)	(0.107)
Job accessibility by transit		0.046	–0.554***	–0.449**
Job accessibility by transit		(0.188)	(0.215)	(0.223)

Note: Only the coefficients of interaction terms between household types and alternative development variables are shown. All control variables were included as in models of Table 2, but are not shown for space reasons. SN = single and no kids; SK = single and with kids; MN = married and no kids; MK = married and with kids.

p < .1. **p < .05. ***p < .01.

Generation, Age, and Income Effects

Table 4 tests generation effects on location preference as many planners have been speculating that Millennials make up a major portion of the market for compact, mixed-use, and transit-oriented development (Millsap 2018; Walker 2016a). The analysis shows mixed results. Millennial households show statistically significantly stronger likelihood to favor transit access in residential location choice than previous generations only when they do not have children. Consumption amenities are a more significant location factor to Millennials than to Gen Xers only in the single with no kids group. However, as they move on to the next stage in life cycle, their preference for these neighborhood attributes turns little different from that of previous generations. We also find that there is no significant difference in location choice in regard to compact, mixed-use, and transit-oriented development between Generation Xers and Baby Boomers.

Table 4.

Heterogeneity in Preference for Alternative Development Across Generations.

	Model 1SN	Model 2SK	Model 3MN	Model 4MK
Reference group—Generation X
Log (distance to downtown)	0.52***	1.15***	1.05***	0.98***
	(0.13)	(0.15)	(0.15)	(0.16)
Compactness index	–0.003	0.002	0.002	–0.004
	(0.003)	(0.003)	(0.003)	(0.003)
Amenities index	0.17***	0.13**	0.20***	0.12**
	(0.05)	(0.05)	(0.05)	(0.06)
Job accessibility by transit	0.23**	0.32***	0.04	0.20*
	(0.10)	(0.10)	(0.11)	(0.11)
Interacted with Millennials:
Log (distance to downtown) × Millennials	0.21	–0.17	0.02	0.20
	(0.20)	(0.23)	(0.38)	(0.37)
Compactness index × Millennials	0.01	–0.001	0.002	0.001
	(0.01)	(0.01)	(0.01)	(0.01)
Amenities index × Millennials	0.15*	0.04	–0.12	–0.22*
	(0.09)	(0.08)	(0.14)	(0.13)
Job accessibility by transit ×* Millennials	0.47***	–0.02	0.44*	–0.13
	(0.17)	(0.16)	(0.26)	(0.25)
Interacted with Boomers
Log (distance to downtown) × Boomers	0.11	–0.27	0.21	0.18
	(0.21)	(0.27)	(0.24)	(0.26)
Compactness × Boomers	–0.01	–0.01	–0.003	0.0002
	(0.01)	(0.01)	(0.005)	(0.01)
Amenities index × Boomers	–0.05	0.12	–0.12	0.04
	(0.09)	(0.11)	(0.08)	(0.09)
Job accessibility by transit × Boomers	–0.12	–0.18	0.22	–0.10
	(0.16)	(0.19)	(0.17)	(0.19)
Observations	2,000	2,000	2,000	2,000
Log likelihood	–2,880.52	–2,942.67	–2,842.55	–2,834.61
Akaike information criterion	5,801.04	5,925.34	5,725.10	5,709.22

Note: Variables in the control group were included in all models as in Table 2, but are not shown for space reasons. We define Millennials, Generation Xers, and Baby Boomers as those who were born between 1985 and 2000, 1965 and 1984, and 1945 and 1964, respectively. Generation Xers are used as a reference group. SN = single and no kids; SK = single and with kids; MN = married and no kids; MK = married and with kids.

p < .1. **p < .05. ***p < .01.

Age also plays only a minor role when life cycle stage is accounted for. Table 5 shows that the age effect is the most pronounced for single households. Consumption amenities become significantly less important even for single households regardless of the presence of children when the household head’s age is fifty and older. When the single householder is young (twenty-nine and younger), they are significantly more likely to move to high-transit-access and compactly developed neighborhoods than older households of the same type. However, age seems not to matter much for married couples in residential location choice in regard of these neighborhood attributes.

Table 5.

Heterogeneity in Preference for Alternative Development Across Age Groups.

	Model 1SN	Model 2SK	Model 3MN	Model 4MK
Reference group—Age thirty to forty-nine
Log (distance to downtown)	0.55***	1.15***	1.11***	1.04***
Log (distance to downtown)	(0.11)	(0.14)	(0.13)	(0.14)
Compactness index	–0.004	0.002	0.001	–0.002
Compactness index	(0.003)	(0.003)	(0.003)	(0.003)
Amenities index	0.19***	0.18***	0.18***	0.12**
Amenities index	(0.05)	(0.05)	(0.05)	(0.05)
Job accessibility by transit	0.16*	0.32***	0.10	0.23**
Job accessibility by transit	(0.08)	(0.09)	(0.09)	(0.10)
Interacted with HH < 30
Log (distance to downtown) × HH < 30	0.18	–0.25	–0.02	–0.07
Log (distance to downtown) × HH < 30	(0.19)	(0.20)	(0.34)	(0.30)
Compactness index × HH < 30	0.01*	–0.004	0.005	–0.01
Compactness index × HH < 30	(0.01)	(0.005)	(0.01)	(0.01)
Amenities index × HH < 30	0.09	–0.03	–0.12	–0.10
Amenities index × HH < 30	(0.08)	(0.08)	(0.12)	(0.10)
Job accessibility by transit × HH < 30	0.52***	–0.08	0.35	–0.40*
Job accessibility by transit × HH < 30	(0.16)	(0.14)	(0.23)	(0.21)
Interacted with HH ≥ 50:
Log (distance to downtown) × HH ≥ 50	–0.12	0.57	0.25	0.78
Log (distance to downtown) × HH ≥ 50	(0.41)	(0.94)	(0.48)	(0.62)
Compactness index × HH ≥ 50	–0.02	0.04*	–0.003	–0.01
Compactness index × HH ≥ 50	(0.01)	(0.02)	(0.01)	(0.01)
Amenities index × HH ≥ 50	–0.35**	–0.64*	–0.24	0.03
Amenities index × HH ≥ 50	(0.15)	(0.34)	(0.15)	(0.18)
Job accessibility by transit × HH ≥ 50	0.10	–0.71	0.23	–0.32
Job accessibility by transit × HH ≥ 50	(0.30)	(0.66)	(0.36)	(0.47)
Observations	2,000	2,000	2,000	2,000
Log likelihood	2,882.03	2,937.45	2,842.72	–2,829.96
Akaike information criterion	5,804.06	5,914.89	5,725.44	5,699.92

Note: Variables in the control group were included in all models as in Table 2, but are not shown for space reasons. Households with a head who is thirty to forty-nine years old are used as the reference group. SN = single and no kids; SK = single and with kids; MN = married and no kids; MK = married and with kids; HH = households.

p < .1. **p < .05. ***p < .01.

Income is found to be a more important household factor because households’ location decisions are conditional on their financial capacity, as shown in Table 6. While consumption amenities or access to vibrant retail and service businesses attract households with medium- to high-income (especially unmarried) households, low-income households tend to move away from amenity-rich neighborhoods, which is more likely to be a result of financial constraints instead of location preferences. Due to the high cost of residing in amenity-rich neighborhoods, low-income households need to trade off their preferences with other factors.

Table 6.

Heterogeneity in Preference for Alternative Development Across Income Groups.

	Model 1SN	Model 2SK	Model 3MN	Model 4MK
Reference group—Medium-income HH
Log (distance to downtown)	0.67***	1.18***	1.44***	1.25***
Log (distance to downtown)	(0.12)	(0.15)	(0.16)	(0.17)
Compactness index	–0.01	0.003	0.005	–0.004
Compactness index	(0.003)	(0.003)	(0.003)	(0.003)
Amenities index	0.31***	0.29***	0.09	0.07
Amenities index	(0.05)	(0.05)	(0.05)	(0.06)
Job accessibility by transit	0.41***	0.19**	0.22**	0.12
Job accessibility by transit	(0.10)	(0.10)	(0.10)	(0.12)
Interacted with low-income HH
Log (distance to downtown) × Low-income HH	0.07	–0.06	–0.34	–1.14**
Log (distance to downtown) × Low-income HH	(0.18)	(0.22)	(0.33)	(0.52)
Compactness index × Low-income HH	0.01*	–0.004	–0.002	–0.0004
Compactness index × Low-income HH	(0.005)	(0.01)	(0.01)	(0.01)
Amenities index × Low-income HH	–0.31***	–0.41***	0.10	–0.37*
Amenities index × Low-income HH	(0.08)	(0.08)	(0.13)	(0.20)
Job accessibility by transit × Low-income HH	–0.16	0.20	–0.11	–0.64*
Job accessibility by transit × Low-income HH	(0.14)	(0.14)	(0.23)	(0.36)
Interacted with high-income HH
Log (distance to downtown) × High-income HH	–0.90***	–0.85**	–0.84***	–0.32
Log (distance to downtown) × High-income HH	(0.30)	(0.40)	(0.24)	(0.25)
Compactness index × High-income HH	–0.01	–0.01	–0.01**	–0.0003
Compactness index × High-income HH	(0.01)	(0.01)	(0.01)	(0.01)
Amenities index × High-income HH	–0.03	0.08	0.13	0.10
Amenities index × High-income HH	(0.15)	(0.16)	(0.09)	(0.08)
Job accessibility by transit × High-income HH	–0.19	–0.04	–0.10	0.17
Job accessibility by transit × High-income HH	(0.28)	(0.31)	(0.18)	(0.17)
Observations	2,000	2,000	2,000	2,000
Log likelihood	2,881.09	2,923.41	2,837.66	–2,828.47
Akaike information criterion	5,802.17	5,886.83	5,715.32	5,696.95

Note: Variables in the control group were included in all models as in Table 2, but are not shown for space reasons. Medium-income households are used as the reference group. SN = single and no kids; SK = single and with kids; MN = married and no kids; MK = married and with kids; HH = households.

p < .1. **p < .05. ***p < .01.

Another notable variable is the distance to downtown. While medium-income households are moving out of the city, most types of high-income households except married households with children are significantly less inclined to do so. High-income single-person households show even recentralizing tendency.⁵ High-income households may value proximity to employment opportunities and rich urban amenities in Chicago downtown and its surrounding areas because they can afford the more expensive housing in those locations. Low-income households are also less likely to suburbanize than medium-income households in case of married couples with households, but perhaps for a different reason. As shown in Figure 3, low-income households’ residential location choice is highly constrained within the central city, especially Chicago South which is the region’s most disadvantaged neighborhoods. In other words, both high-income and low-income households are less likely to decentralize than medium-income households, but for exactly opposite reasons. They also concentrate into different neighborhoods in Chicago, as a result of and resulting in severe segregation.

Conclusion

This paper examined how households’ location decisions are influenced by various components of smart growth that is proposed to counter auto-oriented sprawl. We advanced existing literature on this topic by using a new household mobility dataset, employing new measures of neighborhood-level urbanism, and testing heterogeneous location preferences for compact, mixed-use, and transit-oriented neighborhoods by life cycle stage, age, income, and generation.

MNL modeling results show that while the Chicago region’s households tend to decentralize, they are more likely to move to the neighborhoods with rich consumption amenities and high transit-based accessibility. Location choices toward strong urban characteristics (despite the prevailing decentralization trend) were found regardless of households’ life cycle stages. Compactness of the neighborhood’s physical environment was not a significant location factor for any type of household. The findings suggest that projects incorporating smart growth elements to promote sustainable urban form would work best when they successfully attract vibrant neighborhood businesses and expand public transit access. For example, the TOD component of the GO TO 2040 comprehensive plan for the Chicago region highlighted both mixed land use and high-quality walking environment to transit stations in addition to residential densities in its design guidelines. It also provides a list of many projects that incorporated all three elements (e.g., City of Elmhurst, City of Evanston; Chicago Metropolitan Agency for Planning 2013).

Despite overall evidence for decentralization and high demand for urban characteristics regardless of generation, we found that household preferences for smart growth neighborhoods and the associated location outcomes are significantly affected by life cycle stage and income level. While single-person households have the strongest preference for consumption amenities and job accessibility by transit, marriage and the presence of children significantly reduces the importance of amenities and transit access in residential location outcomes, respectively. Households’ location decisions are also conditional on their financial capacity. While consumption amenities attract households with medium- to high-income (especially unmarried) households, low-income households tend to move away from amenities perhaps due to the high housing costs of living in high-amenity neighborhoods. Both high- and low-income households are less likely to decentralize than middle-income households, but for opposite reasons. While high-income households value proximity to employment opportunities and urban amenities in Chicago downtown and North Chicago, low-income households’ location outcomes are much constrained to the disadvantaged neighborhoods in Southern parts of Chicago. These results indicate a potential mismatch between location preferences and location outcomes, especially for low-income households, which highlights the needs for planners to expand affordable housing opportunities in amenity-rich and high-transit-access neighborhoods.

This research also contributes to the recent discussion of Millennials’ location choice by showing that Millennials, in general, were not much different from previous generations in their preferences for compact, mixed-use, and transit-oriented neighborhoods when life cycle stage is accounted for. However, certain types of Millennials, especially those who are not married and have no kids, are significantly more likely to be attracted to high-amenity and transit-access areas than previous generations of the same household type. Transit access is highly valued even by married Millennial couples until they get a child.

We acknowledge that there are several limitations in this paper. First, the InfoUSA U.S. Consumer Database used in this study does not include several potential determinants of household location choice, such as race or ethnicity (which is information collected by InfoUSA but which was not purchased for the dataset we analyze), employment, and car ownership. It is also difficult to trace the exact relationship among the members of a household, as well as the reasons why they move with the database. Second, while prior research has identified the relative accuracy of InfoUSA address data, opacity regarding the construction, accuracy, and validity of some of the demographic and economic attributes of the data leave some uncertainty regarding the strength of our findings. Third, our analysis is limited to the sample from a single UA, Chicago. In particular, the City of Chicago, like many other central cities in the Midwest, has been struggling with economic decline, population loss, and segregation. Such a declining economy and population loss may affect households’ location behavior in the city and push them to the suburbs, altering their true preferences for compact, mixed-use, and transit-oriented development. Finally, due to the limitation of the available data years, we were not able to single out generational effects from age effects. The data did not allow us to analyze Generation Xers or Baby Boomers’ location choices over time. Nevertheless, the dataset with detailed mobility information and of a large sample size enabled us to analyze households’ actual location decisions, not their willingness to move or home prices. Combining this non-conventional data source and other innovative approaches to operationalizing neighborhood urbanism, we demonstrated a new way to study the role of compact, mixed-use, and transit-oriented development in household location choice.

Footnotes

Acknowledgements

The authors thank Dr. Yongsung Lee at the University of Hong Kong for generously sharing consumption amenities data.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Bumsoo Lee

Andrew Greenlee

Notes

Author Biographies

Yiyuan Wang is a PhD student in the Department of Urban Design and Planning at the University of Washington, Seattle. He is interested in applying statistical and machine learning models to investigate the impacts of new transportation technologies. His recent research studies the public–private partnership between public transportation agencies and private shared mobility companies.

Bumsoo Lee is an associate professor in the Department of Urban and Regional Planning at the University of Illinois at Urbana–Champaign. His research interests include urban spatial organization, sustainable infrastructure and development, and land use and transportation interactions.

Andrew Greenlee is an associate professor in the Department of Urban and Regional Planning at the University of Illinois at Urbana–Champaign. His work focuses on the interaction of housing policy, community development, and social equity in planning.

References

Alonso

William

. 1964. Location and Land Use: Toward a General Theory of Land Rent. Cambridge: Harvard University Press. doi:10.4159/harvard.9780674730854.

Andrew

Mark

Meen

Geoffrey

. 2006. “Population Structure and Location Choice: A Study of London and South East England.” Papers in Regional Science 85 (3): 401–19. doi:10.1111/j.1435-5957.2006.00092.x.

Axhausen

Kay W.

König

Arnd

Scott

Darren M.

Jürgens

Claudia

. 2004. “Locations, Commitments and Activity Spaces.” In Human Behaviour and Traffic Networks, edited by Schreckenberg

Michael

Selten

Reinhard

, 205–30. Berlin: Springer.

Bartholomew

Keith

Ewing

Reid

. 2011. “Hedonic Price Effects of Pedestrian- and Transit-Oriented Development.” Journal of Planning Literature 26 (1): 18–34. doi:10.1177/0885412210386540.

Basolo

Victoria

Yerena

Anaid

. 2017. “Residential Mobility of Low-Income Subsidized Households: A Synthesis of Explanatory Frameworks.” Housing Studies 32 (6): 841–62. doi:10.1080/02673037.2016.1240762.

Bayoh

Isaac

Irwin

Elena G.

Haab

Timothy

. 2006. “Determinants of Residential Location Choice: How Important Are Local Public Goods in Attracting Homeowners to Central City Locations? ” Journal of Regional Science 46 (1): 97–120. doi:10.1111/j.0022-4146.2006.00434.x.

Ben-Akiva

Moshe

Lerman

Steven R.

1985. “Discrete Choice Analysis: Theory and Application to Travel Demand.” Journal of Business & Economic Statistics 6. doi:10.2307/1391567.

Blumenberg

Evelyn

Ralph

Kelcie

Smart

Michael

Taylor

Brian D.

2016. “Who Knows about Kids These Days? Analyzing the Determinants of Youth and Adult Mobility in the U.S. between 1990 and 2009.” Transportation Research Part A: Policy and Practice 93:39–54. doi:10.1016/j.tra.2016.08.010.

Burchell

Robert W.

Shad

Naveed A.

Listokin

David

Phillips

Hilary

Downs

Anthony

Seskin

Samuel

Davis

Judy S.

Moore

Terry

Helton

David

Gall

Michelle

. 1998. The Costs of Sprawl—Revisited. Washington, DC: The National Academies Press.

10.

Cao

Xinyu.

2008. “Is Alternative Development Undersupplied? Examination of Residential Preferences and Choices of Northern California Movers.” Transportation Research Record: Journal of the Transportation Research Board 2077:97–105. doi:10.3141/2077-13.

11.

Chen

Jason

Chen

Cynthia

Timmermans

Harry

. 2008. “Accessibility Trade-Offs in Household Residential Location Decisions.” Transportation Research Record: Journal of the Transportation Research Board 2077:71–79. doi:10.3141/2077-10.

12.

Chicago Metropolitan Agency for Planning. 2013. “Transit-Oriented Development.” Chicago Metropolitan Agency for Planning. https://www.cmap.illinois.gov/about/2040/supporting-materials/process-archive/strategy-papers/urban-design/tod.

13.

Cho

Eun Joo

Rodriguez

Daniel

Song

Yan

. 2008. “The Role of Employment Subcenters in Residential Location Decisions.” Journal of Transport and Land Use 2:121–51. doi:10.5198/jtlu.v3i1.194.

14.

Colwell

Peter F.

Dehring

Carolyn A.

Turnbull

Geoffrey K.

2002. “Recreation Demand and Residential Location.” Journal of Urban Economics 51 (3): 418–28. doi:10.1006/juec.2001.2251.

15.

Couture

Victor

Handbury

Jessie

. 2017. “Urban Revival in America, 2000 to 2010.” No. w24084. National Bureau of Economic Research. doi:10.3386/w24084.

16.

Davis

Benjamin

Dutzik

Tony

Baxandall

Phineas

. 2012. “Transportation and the New Generation.” Frontier Group U.S. PIRG Education Fund. https://uspirg.org/sites/pirg/files/reports/Transportation%20&%20the%20New%20Generation%20vUS_0.pdf.

17.

Dawkins

Casey J.

2005. “Tiebout Choice and Residential Segregation by Race in U.S. Metropolitan Areas, 1980–2000.” Regional Science and Urban Economics 35 (6): 734–55. doi:10.1016/j.regsciurbeco.2005.01.002.

18.

de Palma

André

Motamedi

Kiarash

Picard

Nathalie

Waddell

Paul

. 2005. “A Model of Residential Location Choice with Endogenous Housing Prices and Traffic for the Paris Region.” European Transport/Trasporti Europei 31:67–82. http://www.openstarts.units.it/dspace/bitstream/10077/5884/1/dePalma_et_al_2_ET31.pdf.

19.

de Palma

André

Picard

Nathalie

Waddell

Paul

. 2007. “Discrete Choice Models with Capacity Constraints: An Empirical Analysis of the Housing Market of the Greater Paris Region.” Journal of Urban Economics 62 (2): 204–30. doi:10.1016/j.jue.2007.02.007.

20.

Debrezion

Ghebreegziabiher

Pels

Eric

Rietveld

Piet

. 2007. “The Impact of Railway Stations on Residential and Commercial Property Value: A Meta-analysis.” The Journal of Real Estate Finance and Economics 35 (2): 161–80. doi:10.1007/s11146-007-9032-z.

21.

Dohrmann

Sylvia

Han

Daifeng

Mohadjer

Leyla

. 2006. “Residential Address Lists vs. Traditional Listing: Enumerating Households and Group Quarters.” Proceedings of the Annual Meeting of the American Statistical Association. https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.587.7107&rep=rep1&type=pdf.

22.

Ellis

Cliff

. 2002. “The New Urbanism: Critiques and Rebuttals.” Journal of Urban Design 7 (3): 261–91. doi:10.1080/1357480022000039330.

23.

Ewing

Reid

Cervero

Robert

. 2010. “Travel and the Built Environment.” Journal of the American Planning Association 76 (3): 265–94. doi:10.1080/01944361003766766.

24.

Ewing

Reid

Hamidi

Shima

. 2014. “Measuring Sprawl 2014.” Smart Growth America 45:51.

25.

Friedman

Joseph

. 1981. “A Conditional Logit Model of the Role of Local Public Services in Residential Choice.” Urban Studies 18:347–58. http://usj.sagepub.com/content/18/3/347.short.

26.

Fulton

William B.

1996. The New Urbanism: Hope or Hype for American Communities? Cambridge: Lincoln Institute of Land Policy.

27.

Gabriel

Stuart A.

Rosenthal

Stuart S.

1989. “Household Location and Race: Estimates of a Multinomial Logit Model.” The Review of Economics and Statistics 71 (2): 240–49. doi:10.2307/1926969.

28.

Glaeser

Edward L.

Kolko

Jed

Saiz

Albert

. 2001. “Consumer City.” Journal of Economic Geography 1 (1): 27–50. doi:10.1093/jeg/1.1.27.

29.

Gordon

Peter

Richardson

Harry W.

1997. “Are Compact Cities a Desirable Planning Goal? ” Journal of the American Planning Association 63 (1): 95–106. doi:10.1080/01944369708975727.

30.

Gordon

Peter

Richardson

Harry W.

1998. “A Critique of New Urbanism.” http://www.petergordon.us/urbanism.html.

31.

Gordon

Peter

Richardson

Harry W.

2001. “The Sprawl Debate: Let Markets Plan.” Publius 31 (3): 131–49.

32.

Gottlieb

Paul D.

Lentnek

Barry

. 2001. “Spatial Mismatch Is Not Always a Central-City Problem: An Analysis of Commuting Behaviour in Cleveland, Ohio, and Its Suburbs.” Urban Studies 38 (7): 1161–86. doi:10.1080/00420980120051701.

33.

Greenlee

Andrew J.

2019. “Assessing the Intersection of Neighborhood Change and Residential Mobility Pathways for the Chicago Metropolitan Area (2006–2015).” Housing Policy Debate 29 (1): 186–212. doi:10.1080/10511482.2018.1476898.

34.

Guo

Jessica Y.

Bhat

Chandra R.

2001. Residential Location Choice Modeling: Accommodating Sociodemographic, School Quality and Accessibility Effects. Austin: The University of Texas at Austin.

35.

Guo

Jessica Y.

Bhat

Chandra R.

2007. “Operationalizing the Concept of Neighborhood: Application to Residential Location Choice Analysis.” Journal of Transport Geography 15 (1): 31–45. doi:10.1016/j.jtrangeo.2005.11.001.

36.

Handy

Susan

Sallis

James F.

Weber

Deanne

Maibach

Hollander

Marla

. 2008. “Is Support for Traditionally Designed Communities Growing? Evidence from Two National Surveys.” Journal of the American Planning Association 74 (2): 209–21. doi:10.1080/01944360802010418.

37.

Hörnsten

Lisa

Fredman

Peter

. 2000. “On the Distance to Recreational Forests in Sweden.” Landscape and Urban Planning 51 (1): 1–10. doi:10.1016/S0169-2046(00)00097-9.

38.

Howell-Moroney

Michael

. 2008. “The Tiebout Hypothesis 50 Years Later: Lessons and Lingering Challenges for Metropolitan Governance in the 21st Century.” Public Administration Review 68 (1): 97–109. doi:10.1111/j.1540-6210.2007.00840.x.

39.

Lingqian

Wang

Liming

. 2019. “Housing Location Choices of the Poor: Does Access to Jobs Matter?” Housing Studies 34 (10): 1721–45.

40.

Iannacchione

Vincent

Morton

Katherine

McMichael

Joseph

Cunningham

David

Cajka

James

Chromy

James

. 2007. “Comparing the Coverage of a Household Sampling Frame Based on Mailing Addresses to a Frame Based on Field Enumeration.” Proceedings of the Annual Meeting of the American Statistical Association. http://www.asasrms.org/Proceedings/y2007/Files/JSM2007-000675.pdf.

41.

Kalinoski

Gail

. 2018. “New Urbanism’s Next Chapter.” https://www.multihousingnews.com/post/new-urbanisms-next-chapter-2/.

42.

Kaplan

Rachel

Austin

Maureen E.

2004. “Out in the Country: Sprawl and the Quest for Nature Nearby.” Landscape and Urban Planning 69 (2): 235–43. doi:10.1016/j.landurbplan.2003.09.006.

43.

Kennel

Timothy L.

Mei

. 2009. “Content and Coverage Quality of a Commercial Address List as a National Sampling Frame for Household Surveys.” In Proceedings of the Joint Statistical Meetings 2009, 2364–78. http://www.asasrms.org/Proceedings/y2009/Files/304010.pdf.

44.

Kim

Jae Hong

Pagliara

Francesca

Preston

John

. 2005. “The Intention to Move and Residential Location Choice Behaviour.” Urban Studies 42 (9): 1621–36. doi:10.1080/00420980500185611.

45.

Kim

Junghwan

Lee

Bumsoo

. 2019. “More than Travel Time: New Accessibility Index Capturing the Connectivity of Transit Services.” Journal of Transport Geography 78:8–18. doi:10.1016/201905008.

46.

Kim

Tae-Kyung

Horner

Mark W.

2003. “Exploring Spatial Effects on Urban Housing Duration.” Environment and Planning A 35 (8): 1415–29. doi:10.1068/a35170.

47.

Kim

Tae-Kyung

Horner

Mark W.

Marans

Robert W.

2005. “Life Cycle and Environmental Factors in Selecting Residential and Job Locations.” Housing Studies 20 (3): 457–73. doi:10.1080/02673030500062335.

48.

Kleit

Rachel Garshick

Galvez

Martha

. 2011. “The Location Choices of Public Housing Residents Displaced by Redevelopment: Market Constraints, Personal Preferences, or Social Information? ” Journal of Urban Affairs 334:375–407. doi:10.1111/j.1467-9906.2011.00557.x.

49.

Knaap

Gerrit

Talen

Emily

. 2005. “New Urbanism and Smart Growth: A Few Words from the Academy.” International Regional Science Review 28 (2): 107–18. doi:10.1177/0160017604273621.

50.

Lee

Brian

Paul

H. Y.

Liming Wang

Waddell

Pendyala

Ram M.

2010. “Reexamining the Influence of Work and Nonwork Accessibility on Residential Location Choices with a Microanalytic Framework.” Environment and Planning A 42 (4): 913–30. doi:10.1068/a4291.

51.

Lee

Hyojung

. 2020. “Are Millennials Coming to Town? Residential Location Choice of Young Adults.” Urban Affairs Review 56:565–604. doi:10.1177/1078087418787668.

52.

Lee

Yongsung

Lee

Bumsoo

Shubho

Md Tanvir Hossain

. 2019. “Urban Revival by Millennials? Intraurban Net Migration Patterns of Young Adults, 1980–2010.” Journal of Regional Science 59 (3): 538–66. doi:10.1111/jors.12445.

53.

Levine

Jonathan

. 2005. Zoned Out: Regulation, Markets, and Choices in Transportation and Metropolitan Land-Use. Washington, DC: Resources for the Future.

54.

Levine

Jonathan

Inam

Aseem

. 2004. “The Market for Transportation–Land Use Integration: Do Developers Want Smarter Growth than Regulations Allow?” Transportation 31:409–27. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.584.6966&rep=rep1&type=pdf.

55.

Levine

Jonathan

Inam

Aseem

Torng

Gwo-Wei

. 2005. “A Choice-Based Rationale for Land Use and Transportation Alternatives: Evidence from Boston and Atlanta.” Journal of Planning Education and Research 24 (3): 317–30. doi:10.1177/0739456X04267714.

56.

Lindstrom

Bonnie

. 1997. “A Sense of Place: Housing Selection on Chicago’s North Shore.” The Sociological Quarterly 38 (1): 19–39. http://www.jstor.org/stable/4121264.

57.

Loomis

John B.

2014. “2013 WAEA Keynote Address: Strategies for Overcoming Hypothetical Bias in Stated Preference Surveys.” Journal of Agricultural and Resource Economics 39 (1): 34–46. http://www.jstor.org/stable/44131313.

58.

Manville

Michael

King

David A.

Smart

Michael J.

2017. “The Driving Downturn: A Preliminary Assessment.” Journal of the American Planning Association 83 (1): 42–55. doi:10.1080/01944363.2016.1247653.

59.

Marois

Guillaume

Lord

Sébastien

Negron-Poblete

Paula

. 2018. “The Residential Mobility of Seniors among Different Residential Forms: Analysis of Metropolitan and Rural Issues for Six Contrasted Regions in Québec, Canada.” Journal of Housing for the Elderly 32 (1): 73–98. doi:10.1080/02763893.2017.1393488.

60.

McDonald

Noreen C.

2015. “Are Millennials Really the ‘Go-Nowhere’ Generation? ” Journal of the American Planning Association 81 (2): 90–103. doi:10.1080/01944363.2015.1057196.

61.

McFadden

Daniel Little

. 1978. “Modeling the Choice of Residential Location.” Transportation Research Record: Journal of the Transportation Research Board 673:72–77. http://onlinepubs.trb.org/Onlinepubs/trr/1978/673/673-012.pdf.

62.

Mills

Edwin S.

1967. “An Aggregative Model of Resource Allocation in a Metropolitan Area.” The American Economic Review 57 (2): 197–210. http://www.jstor.org/stable/1821621.

63.

Millsap

Adam

. 2018. “Location Choice in Early Adulthood: Millennials versus Baby Boomers.” Papers in Regional Science 97 (S1): S139–67. doi:10.1111/pirs.12240.

64.

Montgomery

Melanie

Curtis

Carey

. 2006. “Housing Mobility and Location Choice: A Review of the Literature.” Working paper, Urbanet, Department of Urban and Regional Planning, Curtin University. https://pdfs.semanticscholar.org/c2e5/dde3a30e8943aac594835b8e1e70bbd728a3.pdf.

65.

Moos

Markus

. 2014. “‘Generationed’ Space: Societal Restructuring and Young Adults’ Changing Residential Location Patterns.” The Canadian Geographer/Le Géographe canadien 58 (1): 11–33. doi:10.1111/j.1541-0064.2013.12052.x.

66.

Muth

Richard F.

1975. “Numerical Solution of Urban Residential Land-Use Models.” Journal of Urban Economics 2 (4): 307–32. doi:10.1016/0094-1190(75)90031-5.

67.

Myers

Dowell

Gearin

Elizabeth

. 2001. “Current Preferences and Future Demand for Denser Residential Environments.” Housing Policy Debate 12 (4): 633–59. doi:10.1080/10511482.2001.9521422.

68.

National Association of Realtors. 2015. “Community & Transportation Preferences Survey. https://www.nar.realtor/sites/default/files/reports/2015/nar-psu-2015-poll-report.pdf.

69.

O’Muircheartaigh

Colm

Eckman

Stephanie

Weiss

Charlene

. 2002. “Traditional and Enhanced Field Listing for Probability Sampling.” Proceedings of the Annual Meeting of the American Statistical Association. http://www.asasrms.org/Proceedings/y2002/Files/JSM2002-001047.pdf.

70.

O’Muircheartaigh

Colm

English

Ned

Eckman

Stephanie

Upchurch

Heidi

Garcia

Erika

Lepkowski

James

. 2006. “Validating a Sampling Revolution: Benchmarking Address Lists against Traditional Listing.” Proceedings of the Annual Meeting of the American Statistical Association. http://www.asasrms.org/Proceedings/y2006/Files/JSM2006-000909.pdf.

71.

Pinjari

Abdul Rawoof

Pendyala

Ram M.

Bhat

Chandra R.

Waddell

Paul A.

2011. “Modeling the Choice Continuum: An Integrated Model of Residential Location, Auto Ownership, Bicycle Ownership, and Commute Tour Mode Choice Decisions.” Transportation 38 (6): 933–58. doi:10.1007/s11116-011-9360-y.

72.

Reschovsky

Andrew

. 1979. “Residential Choice and the Local Public Sector: An Alternative Test of the ‘Tiebout Hypothesis.’” Journal of Urban Economics 6 (4): 501–20. doi:10.1016/0094-1190(79)90027-5.

73.

Schirmer

Patrick M.

Van Eggermond

Michael A. B.

Axhausen

Kay W.

2014. “The Role of Location in Residential Location Choice Models: A Review of Literature.” Journal of Transport and Land Use 7 (2): 3–21. doi:10.5198/jtlu.v7i2.740.

74.

Shen

Qing

. 1998. “Location Characteristics of Inner-City Neighborhood and Employment Accessibility of Low Wage Workers.” Environment and Planning B 25 (1): 345–65. doi:10.1068/b250345.

75.

Sohmer

Rebecca R.

Lang

Robert E.

2000. “From Seaside to Southside: New Urbanism’s Quest to Save the Inner City.” Housing Policy Debate 11 (4): 751–60. doi:10.1080/10511482.2000.9521386.

76.

Song

Yan

Knaap

Gerrit-Jan

. 2003. “New Urbanism and Housing Values: A Disaggregate Assessment.” Journal of Urban Economics 54 (2): 218–38. doi:10.1016/S0094-1190(03)00059-7.

77.

Song

Yan

Knaap

Gerrit-Jan

. 2004. “Measuring Urban Form: Is Portland Winning the War on Sprawl? ” Journal of the American Planning Association 70 (2): 210–25. doi:10.1080/01944360408976371.

78.

Talen

Emily

. 2000. “New Urbanism and the Culture of Criticism.” Urban Geography 21 (4): 318–41. doi:10.2747/0272-3638.21.4.318.

79.

Tiebout

Charles M.

1956. “A Pure Theory of Local Expenditures.” The Journal of Political Economy 64 (5): 416–24. doi:10.1086/257839.

80.

Tremoulet

Andrée

Dann

Ryan J.

Adkins

Arlie

. 2016. “Moving to Location Affordability? Housing Choice Vouchers and Residential Relocation in the Portland, Oregon, Region.” Housing Policy Debate 26 (4–5): 692–713. doi:10.1080/10511482.2016.1150314.

81.

Trudeau

Dan

. 2013. “A Typology of New Urbanism Neighborhoods.” Journal of Urbanism: International Research on Placemaking and Urban Sustainability 6 (2): 113–38. doi:10.1080/17549175.2013.771695.

82.

Trudeau

Dan

Malloy

Patrick

. 2011. “Suburbs in Disguise? Examining the Geographies of the New Urbanism.” Urban Geography 32 (3): 424–47. doi:10.2747/0272-3638.32.3.424.

83.

Urban Land Institute. 2013. “America in 2013: A ULI Survey of Views on Housing, Transportation, and Community.” https://americas.uli.org/america-2013-uli-survey-views-housing-transportation-community/.

84.

Walker

Alissa

. 2016a. “Millennials Will Live in Cities Unlike Anything We’ve Ever Seen Before.” Gizmodo, April 29. https://gizmodo.com/millennials-will-live-in-cities-unlike-anything-weve-se-1716074100.

85.

Walker

Kyle E.

2016b. “Baby Boomer Migration and Demographic Change in U.S. Metropolitan Areas.” Migration Studies 4 (3): 347–72. doi:10.1093/migration/mnw012.

86.

Weisbrod

Glen

Lerman

Steven R.

Ben-Akiva

Moshe

. 1980. “Tradeoffs in Residential Location Decisions: Transportation versus Other Factors.” Transportation Policy and Decision-Making 1 (1): 13–26.

87.

Zhong

Laitian

Lee

Bumsoo

. 2017. “Carless or Car Later?” Transportation Research Record: Journal of the Transportation Research Board 2664:69–78. doi:10.3141/2664-08.

88.

Zhou

Bin

Kockelman

Kara M.

2008. “Microsimulation of Residential Land Development and Household Location Choices: Bidding for Land in Austin, Texas.” Transportation Research Record: Journal of the Transportation Research Board 2077 (1): 106–12. doi:10.3141/2077-14.

89.

Zondag

Barry

Pieters

Marits

. 2005. “Influence of Accessibility on Residential Location Choice.” Transportation Research Record: Journal of the Transportation Research Board 1902:63–70. doi:10.3141/1902-08.