Do Urban Design qualities add to property values? An empirical analysis of the relationship between Urban Design qualities and property values

Shima Hamidi; Ahmad Bonakdar; Golnaz Keshavarzi; Reid Ewing

doi:10.1016/j.cities.2019.102564

Do Urban Design qualities add to property values? An empirical analysis of the relationship between Urban Design qualities and property values

Shima Hamidi, Ahmad Bonakdar, Golnaz Keshavarzi, Reid Ewing

Bloomberg School of Public Health

Research output: Contribution to journal › Article › peer-review

2 Scopus citations

Abstract

Urban design qualities have the potential to contribute to the sense of safety, comfort, engagement, and overall neighborhood satisfaction perceived by residents, thus can be related to higher property values. Yet, many of these urban design qualities are highly conceptual, require extensive data collection and are subject to various interpretations. As a result, there is little empirical evidence on how street-level urban design qualities are related to property values. Drawing on Multi-level Modeling, this article employs a highly cited dataset on urban design qualities in New York City to provide a statistical analysis of the direct relationship between these qualities and property values. Controlling for confounding factors, this article identifies “imageability” in the street-level environment as a featured urban design quality with the most statistically significant association with property values. In addition, research findings suggest that “transparency” of building facades are positive predictors of property values, whereas the complexity of the built environment exhibits a negative correlation with property values. Policy implications for planners, urban designers, and developers include designing guidelines and street layouts that encourage memorable civic image and identity. While investing in building facades with greater transparency would yield higher property values, complexity in the urban environment, particularly in neighborhoods seeking investment and attracting capital should be treated with care by all parties involved.

Original language	English (US)
Article number	102564
Journal	Cities
Volume	98
DOIs	https://doi.org/10.1016/j.cities.2019.102564
State	Published - Mar 2020

Keywords

Complexity
Enclosure
Imageability
Property values
Residential design premiums
Transparency
Urban design

ASJC Scopus subject areas

Development
Sociology and Political Science
Urban Studies
Tourism, Leisure and Hospitality Management

Access to Document

10.1016/j.cities.2019.102564

Cite this

@article{43c07571183843cf9e8d00c8219db6d4,

title = "Do Urban Design qualities add to property values? An empirical analysis of the relationship between Urban Design qualities and property values",

abstract = "Urban design qualities have the potential to contribute to the sense of safety, comfort, engagement, and overall neighborhood satisfaction perceived by residents, thus can be related to higher property values. Yet, many of these urban design qualities are highly conceptual, require extensive data collection and are subject to various interpretations. As a result, there is little empirical evidence on how street-level urban design qualities are related to property values. Drawing on Multi-level Modeling, this article employs a highly cited dataset on urban design qualities in New York City to provide a statistical analysis of the direct relationship between these qualities and property values. Controlling for confounding factors, this article identifies “imageability” in the street-level environment as a featured urban design quality with the most statistically significant association with property values. In addition, research findings suggest that “transparency” of building facades are positive predictors of property values, whereas the complexity of the built environment exhibits a negative correlation with property values. Policy implications for planners, urban designers, and developers include designing guidelines and street layouts that encourage memorable civic image and identity. While investing in building facades with greater transparency would yield higher property values, complexity in the urban environment, particularly in neighborhoods seeking investment and attracting capital should be treated with care by all parties involved.",

keywords = "Complexity, Enclosure, Imageability, Property values, Residential design premiums, Transparency, Urban design",

author = "Shima Hamidi and Ahmad Bonakdar and Golnaz Keshavarzi and Reid Ewing",

note = "Funding Information: Scholars have long argued that urban design qualities play a profound role in the liveability of residential spaces ( Bacon, 1974 ; Cullen, 1961 ; Jacobs, 1961 ; Jacobs & Appleyard, 1987 ; Lynch, 1960, 1981 ; Whyte, 1980 ). A voluminous body of literature suggests that these qualities often include perceptual characteristics of the built environment such as coherence, comfort, imageability, intricacy, openness, vitality, linkage, transparency, and numerous others. Many of these qualities have the potential to impact place diversity ( Talen, 2006 ), visual aesthetics ( Nasar, 1994 ), sense of comfort and safety ( Dumbaugh & Rae, 2009 ; Stamps, 2005 ), and the level of physical activity ( Badland & Schofield, 2005 ). While urban design qualities could have a diverse range of social and economic implications, some scholars have argued that these qualities may act as psychologically stimulant (or deterrent) forces with regard to individuals' behavior toward the urban environment ( Ewing & Handy, 2009 ; Kaplan & Kaplan, 1989 ; Lynch, 1960 ). Still, little is known about the empirical relationship between urban design qualities and property values, particularly given the fact that individuals' choice of living or working could be dependent upon the quality of street design and buildings' structure. The existing literature identifies the numerous determinants of property values including the physical features of properties such as architectural excellence ( Ahlfeldt & Mastro, 2012 ; Hough & Kratz, 1983 ; Vandell & Lane, 1989 ), as well as the characteristics of the built environment including walkability ( Diao & Ferreira, 2010 ; Gilderbloom, Riggs, & Meares, 2015 ; Pivo & Fisher, 2011 ) or street layout ( Matthews & Turnbull, 2007 ; Ryan & Weber, 2007 ; Song & Knaap, 2003 ). Studies, albeit very limited, have also emphasized the role of visual aesthetics in a range of economic and social outcomes including neighborhood satisfaction ( Florida, Mellander, & Stolarick, 2011 ), and the location decisions of creative entrepreneurs ( Smit, 2011 ). However, despite such efforts that address a few design features related to the urban form (e.g., street layout), further research is needed to shed light on the direct relationship between urban design qualities as manifest in imageability and transparency, among other qualities. Research is limited in providing empirical evidence to what extent street-level urban design qualities might be associated with property values, given the fact that urban design qualities are highly conceptual and subject to various interpretations by scholars and urban design professionals, and therefore, very few efforts have been made to measure them quantitatively. In particular, what makes research even more challenging on a large scale is the nature of urban design qualities that, unlike land use and other neighborhood characteristics that could be easily measured with already available data, require extensive on-site data collection to measure and quantify them. This study seeks to address these gaps in the literature by investigating the relationship between street-level urban design qualities and property values for 303 street segments in New York City (NYC). Unlike most American cities, NYC is uniquely pedestrian-friendly, and is visually reminiscent of major European cities (e.g., Paris, Barcelona, Amsterdam) where children, young people, and older adults engage in walking on a daily basis. The aim of this study, however, is not to test “willingness to pay” and the research findings do not indicate whether, and to what extent, the buyers are willing to pay for higher-priced housing due to better urban design qualities; rather the authors attempt to provide better understanding for municipal planners and city officials on the importance of perceptual urban design qualities beyond walkability. This study employs urban design measures from a series of research studies conducted by Ewing and colleagues (2013; 2009; 2005) that operationalized and quantitatively measured five urban design qualities –imageability, enclosure, human scale, transparency, and complexity. Ewing's team define “imageability” as the quality that makes a place memorable and more recognizable. They refer to “enclosure” as the degree to which buildings visually define streets and public spaces. Similarly, “human scale” denotes the size of physical components that correspond to the size or proportions of humans. They conceptualized “transparency” as the visual permeability of the immediate area beyond the edge of a given street. Finally, they define “complexity” as the ways in which buildings and the general physical environment add to the visual richness of a location. In the following passage, the authors provide a brief overview of the determinants of property values with regards to the physical features of properties. Next, in order to provide a broader context for the built environmental correlates of property values, the authors discuss what elements of the built environment are most often found to be associated with property values, followed by the research hypothesis. 1. Property Values and Urban Design Qualities 1.1. Physical Characteristics of Properties Property value has held a particularly sensitive place among planning scholars and professionals, due to the fundamental role it plays in household economics, while generating a significant revenue stream for local governments. A survey of existing literature on property values indicates two streams of research that have addressed the extent to which externalities and internalities associated with design qualities are correlated with property values. The first body of research focuses on the quintessential design characteristics of buildings in terms of their physical, aesthetics, and structural qualities. The second stream, on the other hand, emphasizes the design qualities of the built environment as well as the place amenities (e.g., access to retail, green space, public transit), as central to real estate values. In the following, related empirical studies will be reviewed. The body of literature on the physical features of property values is well developed and has identified, among other determinants, the physical characteristics of the site such as lot size and building age ( Follain & Jimenez, 1985 ; Peiser, 1987 ), property amenities such as the number of bedrooms, baths, pools ( Waddell, Bemy, & Hoch, 1993 ), architectural significance ( Hough & Kratz, 1983 ; Vandell & Lane, 1989 ), and the design quality of buildings and construction material ( Baum, 1994 ; Knaap, 1998 ; Nase, Berry, & Adair, 2013b, 2016 ). These characteristics are not limited to the US, but also manifest themselves in European housing market. For instance, some studies have found that the high design quality of the building facades demands higher premiums ( Baum, 1993 ; Nase, Berry, & Adair, 2013). The majority of these studies have drawn on hedonic pricing models and reported often consistent findings. 1.2. The Urban Design Qualities and Place Amenities With respect to urban design qualities, a large body of literature suggests the effect of streets with access to retail, ample green space, and mix of land uses on property values. While in part depending on the socio-economic characteristics of a neighborhood (e.g., income, level of education), the evidence suggests potentially high demand for walkable neighborhoods in both residential and non-residential areas ( Diao & Ferreira, 2010 ; Gilderbloom et al., 2015 ; Pivo & Fisher, 2011 ). For instance, using Walk Score to measure walkability, a recent national study on the effect of walkability on the investment returns of 4237 office, apartment, retail and industrial properties (from 2001 to 2008) in the US found that increasing walkability by 10% could lead to an increase of 1–9% in property values, depending on the type of property ( Pivo & Fisher, 2011 ). Similar findings by Diao and Ferreria (2010) in the Boston metropolitan area indicated that one standard deviation increase in walkability score could increase property values by 1.42%, meaning a $5340 increase in a house priced at $376,500. Street layout also tends to correlate with property values ( Asabere, 1990 ; Guttery, 2002 ; Matthews & Turnbull, 2007 ; Song & Knaap, 2003 ). In one study, Matthews and Turnbull (2007) concluded that grid street patterns are of high significance for consumers and positively associated with higher property values, particularly in pedestrian-oriented neighborhoods. Higher property values in this context generally stem from greater accessibility, street connectivity, and mixed land uses exemplified in neotraditionalist neighborhoods ( Koster & Rouwendal, 2012 ; Ryan & Weber, 2007 ; Song & Knaap, 2003 ; Tu & Eppli, 1999 ). Regarding place amenities, studies have generally outlined the positive relationship between higher property values and greater access to neighborhood and street amenities such as public transit (e.g., Bartholomew & Ewing, 2011 ; Bowes & Ihlanfeldt, 2001; Duncan, 2011 ; Hamidi, Kittrell, & Ewing, 2016 ; Hess & Almeida, 2007 ; Mohammad, Graham, Melo, & Anderson, 2013; Nelson et al., 2015 ), public arts ( Roberts & Marsh, 1995 ), parks ( Panduro & Veie, 2013 ), and open spaces ( Irwin, 2002 ; Nilsson, 2014 ). For example, past research has found prices for properties within a quarter-mile of a transit station to be as much as 16.4% higher than those situated farther away from a station ( Debrezion, Pels, & Rietveld, 2007 ). Similarly, using a meta-analysis, another study found that the average single-family home premium was 2.3% higher in transit-adjacent areas than in areas without access to transit ( Hamidi et al., 2016 ). An empirical investigation on the relationship between location-specific amenities such as open landscape in J{\"o}nk{\"o}ping region, Sweden, illustrated that open space amenities in areas with higher density seem to be an important determinant of urban housing market ( Nilsson, 2014 ). Another locational factor is proximity to the Central Business District (CBD) ( Colwell & Munneke, 1997, 1999 ), where proximity to the CBD is generally associated with higher property values. Finally, while architectural design is subjective to individuals' preferences, studies have reported that residents are willing to pay for buildings in proximity to iconic structures of architectural excellence ( Ahlfeldt & Mastro, 2012 ; Buitelaar & Schilder, 2017 ). In a case study of housing market in Netherlands, Buitelaar and Schilder (2017) examined over 60,000 transactions between 1995 and 2014 using a hedonic price model. Their findings demonstrated that vintage buildings attributed to neotraditional styles commanded a price premium of 15%. 2. Research Hypothesis As the existing literature suggests, street-level urban design qualities theoretically correlate with property values. While studies have discussed at length how access to transit, place amenities, and walkability are being capitalized into rising property values, existing literature remain limited in connection with specific urban design qualities as manifest in imageability and transparency, among others. In addition, while urban design qualities have the potential to encourage (or hinder) pedestrian activity and facilitate accessibility to urban amenities and public places, as far as the authors are concerned, research has overlooked the possible, direct association between design qualities related to the built environment and property values. The authors argue that each quality has intrinsic value – offering visual aesthetics, a memorable civic image, a sense of safety and comfort, and a high level of interest and engagement – all of which contribute to neighborhood satisfaction as perceived by residents. Therefore, the findings of this study should be of interest, in particular, to municipal planners, urban designers, developers, and local governments. Drawing on the five urban design qualities identified and operationalized by Ewing and colleagues (2013; 2009; 2005), this study hypothesizes how each quality is most likely related to property values. Imageability is a quality that makes a place vividly identifiable and memorable to visitors, while providing distinguishing features such as color or shape that imprint a mental image on observers ( Lynch, 1960 ). Landmarks and historic buildings are overwhelmingly identified as memorable structures that enhance the imageability of the built environment ( Appleyard, 1969 ). These structures could contribute to the sense of place, and improve psychological well-being of the residents, which would ultimately make a place more appealing for long-term residence ( Nasar, 1990 ). This appeal increases the demand for memorable and identifiable neighborhoods. Therefore, the authors hypothesize that imageability could be positively associated with property values. Enclosure has received mixed reviews in the literature. Existing literature argues that a certain degree of enclosure is a defining factor of “space” within a built boundary or physical contiguity, separating it from other surrounding spaces ( Norberg-Schulz, 1980 ). Some studies have found that visual enclosure is strongly associated with pedestrian walking ( Yin & Wang, 2016 ), whereas some others did not find any significant relationship ( Ewing, Hajrasouliha, Neckerman, Purciel, & Greene, 2016 ). While openness may affect the visual quality of neighborhood, it can be argued that enclosure is generally viewed as opposed to the openness of space, denoting vistas with open views ( Nasar, 1990 ). With regards to safety, landscape views, and overall neighborhood satisfaction, it seems to be the case that individuals prefer not to have much enclosed spaces as they limit the openness of the built environment ( Hur, Nasar, & Chun, 2010 ; Stamps, 2005 ). While existing literature does not clearly address what degree of enclosure is deemed desirable, it can be hypothesized that enclosure could be negatively correlated with property values. Human scale is often considered as the proportion of the built environment elements (e.g., buildings, streets) that correspond to the proportions and scale of the human body. For example, scholars believe that considering human scale in designing street patterns promotes a sense of comfort and pedestrian-scale activity, thus increasing urban charm and “nourishing the human spirit ( Gehl, 1987 ; Kunstler, 1996 ).” Generally, the authors hypothesize that human scale is a desired feature for residents and could be positively correlated with property values. Transparency is one of the most significant and tangible features that urban designers and planners have long incorporated into design guidelines and zoning regulations. It denotes the degree to which an immediate area behind a street's edge is perceptible to the naked eye. Transparent fa{\c c}ades on buildings are suggestive of what Jane Jacobs (1961) refers to as “buzzing street life.” In mixed land use neighborhoods with storefronts, transparent walls and windows provide a welcoming experience for street patrons as well as natural surveillance or more”eyes on the street.” Therefore, there would be an increasing demand for neighborhoods with greater transparency, and the authors hypothesize that transparency could be positively linked to property values. Finally, complexity has received mixed reviews. It is a quality most often referred to as the level of intricacy and visual richness of the built environment, closely tied to the notion of urban vitality ( Kaplan & Kaplan, 1989 ; Montgomery, 1998 ). While complexity seems to be a favorable item on the list of preferred building aesthetics coveted by individuals ( Nasar, 1994 ), existing studies do not arrive at a consensus on whether it positively contributes to the quality of the built environment. Still, the authors hypothesize that neighborhoods with greater complexity of the urban environment could be sought after by residents because of the greater urban dynamics and vitality, and therefore, complexity could be a positive predictor of property values. 3. Methodology The initial sample in this study, as shown in , includes 3279 properties in the blocks facing 588 randomly selected street segments located in all five boroughs of NYC. In our final model, we eliminated properties with more than one unit and also eliminated properties with missing values and ended up with 1120 properties in the blocks facing 303 randomly selected street segments. Data on urban design quality in the City of New York were combined with property-level data from the NYC Department of Finance, which collects property data in the city and assesses their value. New York City was selected based on its urban form and streetscape patterns, since the purpose of the study demanded a well-established urban setting that enjoys a high level of the urban design qualities found in other cities. The specific spatial qualities of NYC, such as pedestrian-friendly neighborhoods also resemble European cities (e.g., Amsterdam, Paris, Prague) where walking as an essential part of daily activities, was a key indicator. Also, New York City offers an array of physical and visual features characterized by walkable and transit-friendly neighborhoods and urban amenities, with mixed land uses and high densities. At the same time, NYC exhibits a variety of built environment features due to its dissimilar boroughs (e.g., the density of Manhattan versus that of Staten Island), and provides enough variation in terms of urban design qualities (e.g., imageability and transparency) to make it a suitable case for study. Fig. 1 One of this study's methodological contributions is in its research design. The majority of studies modeling property values have relied on hedonic price analysis as a function of both neighborhood-level and property-level characteristics, yet rarely do they utilize a multi-level modeling approach. While property values are the result of a multitude of factors associated with property-level and street-level (neighborhood) characteristics, the latter geographic level could exert considerable influence on the property level, creating interdependency between the two levels. This study employed Multi-level Modeling (MLM), as an appropriate method to overcome the nested structure of property and street level characteristics. 3.1. Data The data principally come from two sources. First, the attributes of properties were obtained from the New York City Department of Finance (DOF) that collects property data in NYC and assesses their values. 1 1 http://www1.nyc.gov/site/finance/taxes/property-assessments.page Accessed February 10, 2018. Consistent with a several studies on the topic (e.g. Gilderbloom et al., 2015 ; Sohn, Moudon, & Lee 2012), this study utilized these assessments as a proxy for property values –this study's dependent variable. Given the fact that this study examines only 588 randomly selected block faces (303 street segments, when counting both sides of the streets), the primary reason for choosing NYC DoF assessor estimates instead of transactions sales data is the limited and very small sample size associated with transactions sales data ( Grout, Jaeger, & Plantinga, 2011 ; Hess & Almeida, 2007 ; Kay, Noland, & DiPetrillo, 2014 ). Our first attempt was to use the real estate transaction data as our dependent variable. We gathered real estate transaction data for every year between 2008 and 2015. We found only 21 transactions in 2015 facing the block in our sample. Similarly, we found only 19 sales in 2014, 18 sales in 2013, and only 9 in 2012. We went back to 2008 and found the same trend between 2008 and 2011. Even if we add the units sold for all these years, it would not give us sufficient cases for the analysis. Moreover, some of these sales happened during the recession (2008–2010) and the post-recession period and not all places were affected by the economic recession in the same way. The use of assessed value allow us to include every property that face the 588 block faces in our sample in our analysis. Previous studies have demonstrated that these estimates are strongly correlated with the actual sales data and, hence, can provide reliable estimates of property prices ( Goodman & Ittner, 1992 ; Hipp & Singh, 2014 ; Kiel & Zabel, 1999 ). Secondly, the authors acquired other data points from the New York City Department of City Planning. These variables were intended to capture the physical characteristics of properties, including building age, building area, and FAR, among other parcel-level features. The authors adopted the urban design data from a study done by Ewing, et al. (2013), which aimed to provide operational measurement protocols for the 5 urban design qualities related to walkability in New York City ( ). As discussed earlier, Ewing et al. (2009; 2005) operationalized these qualities, drawing on an expert panel that rated a series of video clips containing streetscape features. These video clips included 200 plus video clips from different cities across the US, focusing on main street settings with strong urban affinity mostly located in commercial and business districts. The procedure of operationalizing the perceptual urban design qualities by Ewing and his team involved 1) having the expert panel rate urban design qualities of videotaped streetscapes with respect to urban design qualities; 2) measuring physical features of streetscapes from the video-clips; 3) statistically analyzing relationships between physical features and urban design quality ratings by the expert panel; 4) selecting five urban design qualities namely complexity, imageability, transparency, human scale, and enclosure; and 5) developing, testing, and refining a field manual that laid out procedures for measuring these qualities. illustrates examples of high and low scores for each quality associated with random block faces. (See .) Table 1 Fig. 2 Table 2 While Ewing and colleagues' operationalization of perceptual urban design qualities is a novel and pioneering approach, it comes with limitations. Frist of all, their work focused on urban settings and is not generalizable to suburban, exurban, and rural areas. Secondly, the operationalization was done based on ratings of experts in urban design, planning and related field which could be different from the typical users of the street space. Finally, the ratings were done based on limited videotapes and in a lab setting rather than in-field ratings. It is possible that the perceptions in a real environment are more complex. A handful of follow-up studies have used these streetscape features, providing greater and validity ( Ameli, Hamidi, Garfinkel-Castro, & Ewing, 2015 ; Hamidi & Moazzeni, 2019 ; Purciel et al., 2009 ). In one of these studies, funded by the Robert Wood Johnson Foundation, researchers from Columbia University developed observational data for the five urban design measures for a sample of streets in New York City ( Purciel et al., 2009 ). The researchers began by developing a field manual, based on the original field manual ( Clemente et al., 2006 ), but with all photographic images specific to New York City and with more detailed guidance to observers than does the original manual. Next, a team of six trained observers were sent into field to make measurements for a stratified sample of 588 block faces in all five boroughs that comprise New York City. To assess inter-rater reliability, thirteen block faces were scored independently by all observers, and inter-class correlation coefficients were calculated for each urban design quality and each rater. Results indicated a high degree of consistency among field observers ( Purciel et al., 2009 ). These data are used in the current study to explore the relationship between urban design qualities and property value. However, there is a certain degree of overlap between the specific features of each urban design quality. For example, the number of people present is a feature shared between imageability and complexity. Also, the proportion of windows at street level has overlap between both human scale and transparency. In order to control for such a multi-collinearity inherent in some of these constructs, the authors used average values for such features to remove any potential variation between them that causes multi-collinearity. These features include number of buildings with non-rectangular shapes, the presence of outdoor dining, the number of people present, the number of long sight lines, the proportion of the street wall, and the proportions of windows at street level. Using this method, the authors were able to recalculate new scores for each urban design construct based on the original field study manual illustrated by Ewing and his colleagues 2 2 http://smartgrowth.umd.edu/assets/documents/research/ewingclementehandyetal_walkableurbandesign_2005.pdf . (see Formula 1 below and Table 1 for the full name of variables in the formula). The new scores were also tested against the original scores and found to have a great deal of similarity, ensuring their reliability and accuracy. Formula 1: Imageability Score = 2.44 + (i11*0.41) + (i12*0.72) + (i13*0.97) + (i14* 0.11) + (9.524*0.08) + (0.012*0.64) + (5.685*0.02) + (i18*-0.18). Enclosure Score = 2.57 + (0.417*-0.31) + (0.567*0.72) + (e222*0.94) + (e231*-1.42) + (e232*-2.19) Human Scale Score = 2.61 + (0.4*-0.74) + (0.238*1.10) + (h33*-0.003) + (h34*0.05) + (h35*0.04) Transparency Score = 1.71 + (t41*1.22) + (t42*0.67) + (t43*0.53) Complexity Score = 2.61 + (c51*0.05) + (c521*0.23) + (c522*0.12) + (c53*0.42) + (c54*0.29) + (c55*0.03) Where i17 and c53 are dummy variables: i17 = 0 (if outdoor dining is absent); and i17 = 1 (if outdoor dining is present). c53 = 0 ( if outdoor dining is absent); and c53 = 1 (if outdoor dining is present). 3.2. Control Variables Property characteristics were used at the property level, and built environment characteristics were used at the street segment (neighborhood) level as control variables. GIS data for the study area, including MapPluto parcel layers and street segment centerlines, were acquired from the New York City Department of City Planning. MapPluto files contain each lot's tax data, with the lot's physical features stored in ESRI shape file and database table file formats. These data categorized the buildings by use/type, such as office, retail, or residential. At the property level, the authors included building age (BUIDLING AGE), building lot frontage facing the street (LN LOT FRONT), and Floor-Area-Ratio (FAR), capturing the physical characteristics of the properties. The authors also controlled for single-family residential properties using a dummy variable (SINGLE RES). At the street segment level, specific variables were included to represent accessibility to destinations likely to be related to property values. First is the degree of walkability captured by Walk Score{\textregistered}, a web-based platform that collects the number of amenities (e.g., grocery stores, restaurants, schools) within a walkable distance of a specific address in order to rate its walkability on a numeric scale ranging from 0 to 100. Drawing on Walk Score{\textregistered} data, this study used the walk score for each street segment using the physical address of properties in the center of each block face (from an intersection to the next intersection) (WALK SCORE). The next variable included in the model is distance to transit (NET DISTANCE TRANS). Proximity to transit is heavily studied in hedonic models as a significant determinant of property value. Using ESRI ArcInfo Network Analyst and New York City road centerline shape files, the authors measured the shortest distance (in miles) to the closest rail station from each study segment center point. Additionally, the distance to CBD (DISTANCE CBD), also in miles, were measured, since this could be associated with property values. Mixed land use is often considered contributor to pedestrian-friendly development, potentially correlated with property values. Entropy (BLOCK ENTROPY) was used as a measure of mixed land use, in the census block where the property is located, using Formula 2, borrowed from Ewing and Clemente (2013) . Formula 2: entropy = −[residential share*LN (residential share) + retail share*LN (retail share) + office share*LN (office share)]/LN (3). … where the shares were computed based on floor area of each use for lots within the buffer (entropy). 3.3. Analytical Methods The independent variables in this study come from two geographic levels: property level and street level. Since all properties in the same block share the same urban design qualities and the built environmental characteristics (e.g., imageability, distance to rail) of the street segment in which properties are located, a nesting structure tends to emerge. This nesting structure generates correlation between observations, and violates the assumption of independence of errors attributed to Ordinary Least Squares (OLS). In this case, OLS may not accurately reveal the variances and may underestimate the standard errors of regression coefficients. The appropriate statistical method for nested data is Multilevel Modeling (MLM), also known as Hierarchical Modeling (HLM) ( Goldstein, 2011 ; Heck & Thomas, 2015 ). It allows for overcoming the limitations of OLS and controlling for dependence among lower level cases, and hence yields more precise coefficients and standard error estimations. The authors employed HLM 6.08 software to perform the MLM and to control for the dependence of properties within a given block on the street segment characteristics (Formula 3). Formula 3: Level-1 Model. f ( Ln PROPERTY VALUE ) = B 0 + B 1 *(Ln LOT FRONT) + B 2 *(Ln BUILDING AGE) + B 3 *(FAR) + B 4 *(SINGLE RES). Where SINGLE RES is a dummy variable: SINGLE RES = 0 (if the unit is not a single-family). SINGLE RES = 1 (if the unit is a single-family). Level-2 Model. B 0 = G 00 + G 01 *(IMAGEABILITY) + G 02 *(ENCLOSURE) + G 03 *(HUMAN SCALE) + G 04 *(TRANSPARENCY) + G 05 *(COMPLEXITY) + G 06 *(DISTANCE CBD) + G 07 *(BLOCK ENTROPY) + G 08 *(WALK SCORE) + G 09 *(NET DISTANCE TRANS). Initially, the authors ran a correlations matrix to see the potential association between the independent variables in both levels and property values. At level-1, FAR, lot front area are positively associated with property values (Pearson Correlation (PC) 0.611 and 0.615 respectively) and building age is negatively associated with property value with PC value of −0.128. Also, the correlation matrix demonstrates that, at level-2, the distance to transit and also to CBD exhibit negative relationship (PC of −0.224 and − 0.424, respectively) while Walkscore shows a positive association with property values (PC of 0.349). Finally, the authors tested for multi-collinearity using the VIF threshold of 10 following by (Hair et al. 2009). 2 Publisher Copyright: {\textcopyright} 2019 Elsevier Ltd",

year = "2020",

month = mar,

doi = "10.1016/j.cities.2019.102564",

language = "English (US)",

volume = "98",

journal = "Cities",

issn = "0264-2751",

publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Do Urban Design qualities add to property values? An empirical analysis of the relationship between Urban Design qualities and property values

AU - Hamidi, Shima

AU - Bonakdar, Ahmad

AU - Keshavarzi, Golnaz

AU - Ewing, Reid

N1 - Funding Information: Scholars have long argued that urban design qualities play a profound role in the liveability of residential spaces ( Bacon, 1974 ; Cullen, 1961 ; Jacobs, 1961 ; Jacobs & Appleyard, 1987 ; Lynch, 1960, 1981 ; Whyte, 1980 ). A voluminous body of literature suggests that these qualities often include perceptual characteristics of the built environment such as coherence, comfort, imageability, intricacy, openness, vitality, linkage, transparency, and numerous others. Many of these qualities have the potential to impact place diversity ( Talen, 2006 ), visual aesthetics ( Nasar, 1994 ), sense of comfort and safety ( Dumbaugh & Rae, 2009 ; Stamps, 2005 ), and the level of physical activity ( Badland & Schofield, 2005 ). While urban design qualities could have a diverse range of social and economic implications, some scholars have argued that these qualities may act as psychologically stimulant (or deterrent) forces with regard to individuals' behavior toward the urban environment ( Ewing & Handy, 2009 ; Kaplan & Kaplan, 1989 ; Lynch, 1960 ). Still, little is known about the empirical relationship between urban design qualities and property values, particularly given the fact that individuals' choice of living or working could be dependent upon the quality of street design and buildings' structure. The existing literature identifies the numerous determinants of property values including the physical features of properties such as architectural excellence ( Ahlfeldt & Mastro, 2012 ; Hough & Kratz, 1983 ; Vandell & Lane, 1989 ), as well as the characteristics of the built environment including walkability ( Diao & Ferreira, 2010 ; Gilderbloom, Riggs, & Meares, 2015 ; Pivo & Fisher, 2011 ) or street layout ( Matthews & Turnbull, 2007 ; Ryan & Weber, 2007 ; Song & Knaap, 2003 ). Studies, albeit very limited, have also emphasized the role of visual aesthetics in a range of economic and social outcomes including neighborhood satisfaction ( Florida, Mellander, & Stolarick, 2011 ), and the location decisions of creative entrepreneurs ( Smit, 2011 ). However, despite such efforts that address a few design features related to the urban form (e.g., street layout), further research is needed to shed light on the direct relationship between urban design qualities as manifest in imageability and transparency, among other qualities. Research is limited in providing empirical evidence to what extent street-level urban design qualities might be associated with property values, given the fact that urban design qualities are highly conceptual and subject to various interpretations by scholars and urban design professionals, and therefore, very few efforts have been made to measure them quantitatively. In particular, what makes research even more challenging on a large scale is the nature of urban design qualities that, unlike land use and other neighborhood characteristics that could be easily measured with already available data, require extensive on-site data collection to measure and quantify them. This study seeks to address these gaps in the literature by investigating the relationship between street-level urban design qualities and property values for 303 street segments in New York City (NYC). Unlike most American cities, NYC is uniquely pedestrian-friendly, and is visually reminiscent of major European cities (e.g., Paris, Barcelona, Amsterdam) where children, young people, and older adults engage in walking on a daily basis. The aim of this study, however, is not to test “willingness to pay” and the research findings do not indicate whether, and to what extent, the buyers are willing to pay for higher-priced housing due to better urban design qualities; rather the authors attempt to provide better understanding for municipal planners and city officials on the importance of perceptual urban design qualities beyond walkability. This study employs urban design measures from a series of research studies conducted by Ewing and colleagues (2013; 2009; 2005) that operationalized and quantitatively measured five urban design qualities –imageability, enclosure, human scale, transparency, and complexity. Ewing's team define “imageability” as the quality that makes a place memorable and more recognizable. They refer to “enclosure” as the degree to which buildings visually define streets and public spaces. Similarly, “human scale” denotes the size of physical components that correspond to the size or proportions of humans. They conceptualized “transparency” as the visual permeability of the immediate area beyond the edge of a given street. Finally, they define “complexity” as the ways in which buildings and the general physical environment add to the visual richness of a location. In the following passage, the authors provide a brief overview of the determinants of property values with regards to the physical features of properties. Next, in order to provide a broader context for the built environmental correlates of property values, the authors discuss what elements of the built environment are most often found to be associated with property values, followed by the research hypothesis. 1. Property Values and Urban Design Qualities 1.1. Physical Characteristics of Properties Property value has held a particularly sensitive place among planning scholars and professionals, due to the fundamental role it plays in household economics, while generating a significant revenue stream for local governments. A survey of existing literature on property values indicates two streams of research that have addressed the extent to which externalities and internalities associated with design qualities are correlated with property values. The first body of research focuses on the quintessential design characteristics of buildings in terms of their physical, aesthetics, and structural qualities. The second stream, on the other hand, emphasizes the design qualities of the built environment as well as the place amenities (e.g., access to retail, green space, public transit), as central to real estate values. In the following, related empirical studies will be reviewed. The body of literature on the physical features of property values is well developed and has identified, among other determinants, the physical characteristics of the site such as lot size and building age ( Follain & Jimenez, 1985 ; Peiser, 1987 ), property amenities such as the number of bedrooms, baths, pools ( Waddell, Bemy, & Hoch, 1993 ), architectural significance ( Hough & Kratz, 1983 ; Vandell & Lane, 1989 ), and the design quality of buildings and construction material ( Baum, 1994 ; Knaap, 1998 ; Nase, Berry, & Adair, 2013b, 2016 ). These characteristics are not limited to the US, but also manifest themselves in European housing market. For instance, some studies have found that the high design quality of the building facades demands higher premiums ( Baum, 1993 ; Nase, Berry, & Adair, 2013). The majority of these studies have drawn on hedonic pricing models and reported often consistent findings. 1.2. The Urban Design Qualities and Place Amenities With respect to urban design qualities, a large body of literature suggests the effect of streets with access to retail, ample green space, and mix of land uses on property values. While in part depending on the socio-economic characteristics of a neighborhood (e.g., income, level of education), the evidence suggests potentially high demand for walkable neighborhoods in both residential and non-residential areas ( Diao & Ferreira, 2010 ; Gilderbloom et al., 2015 ; Pivo & Fisher, 2011 ). For instance, using Walk Score to measure walkability, a recent national study on the effect of walkability on the investment returns of 4237 office, apartment, retail and industrial properties (from 2001 to 2008) in the US found that increasing walkability by 10% could lead to an increase of 1–9% in property values, depending on the type of property ( Pivo & Fisher, 2011 ). Similar findings by Diao and Ferreria (2010) in the Boston metropolitan area indicated that one standard deviation increase in walkability score could increase property values by 1.42%, meaning a $5340 increase in a house priced at $376,500. Street layout also tends to correlate with property values ( Asabere, 1990 ; Guttery, 2002 ; Matthews & Turnbull, 2007 ; Song & Knaap, 2003 ). In one study, Matthews and Turnbull (2007) concluded that grid street patterns are of high significance for consumers and positively associated with higher property values, particularly in pedestrian-oriented neighborhoods. Higher property values in this context generally stem from greater accessibility, street connectivity, and mixed land uses exemplified in neotraditionalist neighborhoods ( Koster & Rouwendal, 2012 ; Ryan & Weber, 2007 ; Song & Knaap, 2003 ; Tu & Eppli, 1999 ). Regarding place amenities, studies have generally outlined the positive relationship between higher property values and greater access to neighborhood and street amenities such as public transit (e.g., Bartholomew & Ewing, 2011 ; Bowes & Ihlanfeldt, 2001; Duncan, 2011 ; Hamidi, Kittrell, & Ewing, 2016 ; Hess & Almeida, 2007 ; Mohammad, Graham, Melo, & Anderson, 2013; Nelson et al., 2015 ), public arts ( Roberts & Marsh, 1995 ), parks ( Panduro & Veie, 2013 ), and open spaces ( Irwin, 2002 ; Nilsson, 2014 ). For example, past research has found prices for properties within a quarter-mile of a transit station to be as much as 16.4% higher than those situated farther away from a station ( Debrezion, Pels, & Rietveld, 2007 ). Similarly, using a meta-analysis, another study found that the average single-family home premium was 2.3% higher in transit-adjacent areas than in areas without access to transit ( Hamidi et al., 2016 ). An empirical investigation on the relationship between location-specific amenities such as open landscape in Jönköping region, Sweden, illustrated that open space amenities in areas with higher density seem to be an important determinant of urban housing market ( Nilsson, 2014 ). Another locational factor is proximity to the Central Business District (CBD) ( Colwell & Munneke, 1997, 1999 ), where proximity to the CBD is generally associated with higher property values. Finally, while architectural design is subjective to individuals' preferences, studies have reported that residents are willing to pay for buildings in proximity to iconic structures of architectural excellence ( Ahlfeldt & Mastro, 2012 ; Buitelaar & Schilder, 2017 ). In a case study of housing market in Netherlands, Buitelaar and Schilder (2017) examined over 60,000 transactions between 1995 and 2014 using a hedonic price model. Their findings demonstrated that vintage buildings attributed to neotraditional styles commanded a price premium of 15%. 2. Research Hypothesis As the existing literature suggests, street-level urban design qualities theoretically correlate with property values. While studies have discussed at length how access to transit, place amenities, and walkability are being capitalized into rising property values, existing literature remain limited in connection with specific urban design qualities as manifest in imageability and transparency, among others. In addition, while urban design qualities have the potential to encourage (or hinder) pedestrian activity and facilitate accessibility to urban amenities and public places, as far as the authors are concerned, research has overlooked the possible, direct association between design qualities related to the built environment and property values. The authors argue that each quality has intrinsic value – offering visual aesthetics, a memorable civic image, a sense of safety and comfort, and a high level of interest and engagement – all of which contribute to neighborhood satisfaction as perceived by residents. Therefore, the findings of this study should be of interest, in particular, to municipal planners, urban designers, developers, and local governments. Drawing on the five urban design qualities identified and operationalized by Ewing and colleagues (2013; 2009; 2005), this study hypothesizes how each quality is most likely related to property values. Imageability is a quality that makes a place vividly identifiable and memorable to visitors, while providing distinguishing features such as color or shape that imprint a mental image on observers ( Lynch, 1960 ). Landmarks and historic buildings are overwhelmingly identified as memorable structures that enhance the imageability of the built environment ( Appleyard, 1969 ). These structures could contribute to the sense of place, and improve psychological well-being of the residents, which would ultimately make a place more appealing for long-term residence ( Nasar, 1990 ). This appeal increases the demand for memorable and identifiable neighborhoods. Therefore, the authors hypothesize that imageability could be positively associated with property values. Enclosure has received mixed reviews in the literature. Existing literature argues that a certain degree of enclosure is a defining factor of “space” within a built boundary or physical contiguity, separating it from other surrounding spaces ( Norberg-Schulz, 1980 ). Some studies have found that visual enclosure is strongly associated with pedestrian walking ( Yin & Wang, 2016 ), whereas some others did not find any significant relationship ( Ewing, Hajrasouliha, Neckerman, Purciel, & Greene, 2016 ). While openness may affect the visual quality of neighborhood, it can be argued that enclosure is generally viewed as opposed to the openness of space, denoting vistas with open views ( Nasar, 1990 ). With regards to safety, landscape views, and overall neighborhood satisfaction, it seems to be the case that individuals prefer not to have much enclosed spaces as they limit the openness of the built environment ( Hur, Nasar, & Chun, 2010 ; Stamps, 2005 ). While existing literature does not clearly address what degree of enclosure is deemed desirable, it can be hypothesized that enclosure could be negatively correlated with property values. Human scale is often considered as the proportion of the built environment elements (e.g., buildings, streets) that correspond to the proportions and scale of the human body. For example, scholars believe that considering human scale in designing street patterns promotes a sense of comfort and pedestrian-scale activity, thus increasing urban charm and “nourishing the human spirit ( Gehl, 1987 ; Kunstler, 1996 ).” Generally, the authors hypothesize that human scale is a desired feature for residents and could be positively correlated with property values. Transparency is one of the most significant and tangible features that urban designers and planners have long incorporated into design guidelines and zoning regulations. It denotes the degree to which an immediate area behind a street's edge is perceptible to the naked eye. Transparent façades on buildings are suggestive of what Jane Jacobs (1961) refers to as “buzzing street life.” In mixed land use neighborhoods with storefronts, transparent walls and windows provide a welcoming experience for street patrons as well as natural surveillance or more”eyes on the street.” Therefore, there would be an increasing demand for neighborhoods with greater transparency, and the authors hypothesize that transparency could be positively linked to property values. Finally, complexity has received mixed reviews. It is a quality most often referred to as the level of intricacy and visual richness of the built environment, closely tied to the notion of urban vitality ( Kaplan & Kaplan, 1989 ; Montgomery, 1998 ). While complexity seems to be a favorable item on the list of preferred building aesthetics coveted by individuals ( Nasar, 1994 ), existing studies do not arrive at a consensus on whether it positively contributes to the quality of the built environment. Still, the authors hypothesize that neighborhoods with greater complexity of the urban environment could be sought after by residents because of the greater urban dynamics and vitality, and therefore, complexity could be a positive predictor of property values. 3. Methodology The initial sample in this study, as shown in , includes 3279 properties in the blocks facing 588 randomly selected street segments located in all five boroughs of NYC. In our final model, we eliminated properties with more than one unit and also eliminated properties with missing values and ended up with 1120 properties in the blocks facing 303 randomly selected street segments. Data on urban design quality in the City of New York were combined with property-level data from the NYC Department of Finance, which collects property data in the city and assesses their value. New York City was selected based on its urban form and streetscape patterns, since the purpose of the study demanded a well-established urban setting that enjoys a high level of the urban design qualities found in other cities. The specific spatial qualities of NYC, such as pedestrian-friendly neighborhoods also resemble European cities (e.g., Amsterdam, Paris, Prague) where walking as an essential part of daily activities, was a key indicator. Also, New York City offers an array of physical and visual features characterized by walkable and transit-friendly neighborhoods and urban amenities, with mixed land uses and high densities. At the same time, NYC exhibits a variety of built environment features due to its dissimilar boroughs (e.g., the density of Manhattan versus that of Staten Island), and provides enough variation in terms of urban design qualities (e.g., imageability and transparency) to make it a suitable case for study. Fig. 1 One of this study's methodological contributions is in its research design. The majority of studies modeling property values have relied on hedonic price analysis as a function of both neighborhood-level and property-level characteristics, yet rarely do they utilize a multi-level modeling approach. While property values are the result of a multitude of factors associated with property-level and street-level (neighborhood) characteristics, the latter geographic level could exert considerable influence on the property level, creating interdependency between the two levels. This study employed Multi-level Modeling (MLM), as an appropriate method to overcome the nested structure of property and street level characteristics. 3.1. Data The data principally come from two sources. First, the attributes of properties were obtained from the New York City Department of Finance (DOF) that collects property data in NYC and assesses their values. 1 1 http://www1.nyc.gov/site/finance/taxes/property-assessments.page Accessed February 10, 2018. Consistent with a several studies on the topic (e.g. Gilderbloom et al., 2015 ; Sohn, Moudon, & Lee 2012), this study utilized these assessments as a proxy for property values –this study's dependent variable. Given the fact that this study examines only 588 randomly selected block faces (303 street segments, when counting both sides of the streets), the primary reason for choosing NYC DoF assessor estimates instead of transactions sales data is the limited and very small sample size associated with transactions sales data ( Grout, Jaeger, & Plantinga, 2011 ; Hess & Almeida, 2007 ; Kay, Noland, & DiPetrillo, 2014 ). Our first attempt was to use the real estate transaction data as our dependent variable. We gathered real estate transaction data for every year between 2008 and 2015. We found only 21 transactions in 2015 facing the block in our sample. Similarly, we found only 19 sales in 2014, 18 sales in 2013, and only 9 in 2012. We went back to 2008 and found the same trend between 2008 and 2011. Even if we add the units sold for all these years, it would not give us sufficient cases for the analysis. Moreover, some of these sales happened during the recession (2008–2010) and the post-recession period and not all places were affected by the economic recession in the same way. The use of assessed value allow us to include every property that face the 588 block faces in our sample in our analysis. Previous studies have demonstrated that these estimates are strongly correlated with the actual sales data and, hence, can provide reliable estimates of property prices ( Goodman & Ittner, 1992 ; Hipp & Singh, 2014 ; Kiel & Zabel, 1999 ). Secondly, the authors acquired other data points from the New York City Department of City Planning. These variables were intended to capture the physical characteristics of properties, including building age, building area, and FAR, among other parcel-level features. The authors adopted the urban design data from a study done by Ewing, et al. (2013), which aimed to provide operational measurement protocols for the 5 urban design qualities related to walkability in New York City ( ). As discussed earlier, Ewing et al. (2009; 2005) operationalized these qualities, drawing on an expert panel that rated a series of video clips containing streetscape features. These video clips included 200 plus video clips from different cities across the US, focusing on main street settings with strong urban affinity mostly located in commercial and business districts. The procedure of operationalizing the perceptual urban design qualities by Ewing and his team involved 1) having the expert panel rate urban design qualities of videotaped streetscapes with respect to urban design qualities; 2) measuring physical features of streetscapes from the video-clips; 3) statistically analyzing relationships between physical features and urban design quality ratings by the expert panel; 4) selecting five urban design qualities namely complexity, imageability, transparency, human scale, and enclosure; and 5) developing, testing, and refining a field manual that laid out procedures for measuring these qualities. illustrates examples of high and low scores for each quality associated with random block faces. (See .) Table 1 Fig. 2 Table 2 While Ewing and colleagues' operationalization of perceptual urban design qualities is a novel and pioneering approach, it comes with limitations. Frist of all, their work focused on urban settings and is not generalizable to suburban, exurban, and rural areas. Secondly, the operationalization was done based on ratings of experts in urban design, planning and related field which could be different from the typical users of the street space. Finally, the ratings were done based on limited videotapes and in a lab setting rather than in-field ratings. It is possible that the perceptions in a real environment are more complex. A handful of follow-up studies have used these streetscape features, providing greater and validity ( Ameli, Hamidi, Garfinkel-Castro, & Ewing, 2015 ; Hamidi & Moazzeni, 2019 ; Purciel et al., 2009 ). In one of these studies, funded by the Robert Wood Johnson Foundation, researchers from Columbia University developed observational data for the five urban design measures for a sample of streets in New York City ( Purciel et al., 2009 ). The researchers began by developing a field manual, based on the original field manual ( Clemente et al., 2006 ), but with all photographic images specific to New York City and with more detailed guidance to observers than does the original manual. Next, a team of six trained observers were sent into field to make measurements for a stratified sample of 588 block faces in all five boroughs that comprise New York City. To assess inter-rater reliability, thirteen block faces were scored independently by all observers, and inter-class correlation coefficients were calculated for each urban design quality and each rater. Results indicated a high degree of consistency among field observers ( Purciel et al., 2009 ). These data are used in the current study to explore the relationship between urban design qualities and property value. However, there is a certain degree of overlap between the specific features of each urban design quality. For example, the number of people present is a feature shared between imageability and complexity. Also, the proportion of windows at street level has overlap between both human scale and transparency. In order to control for such a multi-collinearity inherent in some of these constructs, the authors used average values for such features to remove any potential variation between them that causes multi-collinearity. These features include number of buildings with non-rectangular shapes, the presence of outdoor dining, the number of people present, the number of long sight lines, the proportion of the street wall, and the proportions of windows at street level. Using this method, the authors were able to recalculate new scores for each urban design construct based on the original field study manual illustrated by Ewing and his colleagues 2 2 http://smartgrowth.umd.edu/assets/documents/research/ewingclementehandyetal_walkableurbandesign_2005.pdf . (see Formula 1 below and Table 1 for the full name of variables in the formula). The new scores were also tested against the original scores and found to have a great deal of similarity, ensuring their reliability and accuracy. Formula 1: Imageability Score = 2.44 + (i11*0.41) + (i12*0.72) + (i13*0.97) + (i14* 0.11) + (9.524*0.08) + (0.012*0.64) + (5.685*0.02) + (i18*-0.18). Enclosure Score = 2.57 + (0.417*-0.31) + (0.567*0.72) + (e222*0.94) + (e231*-1.42) + (e232*-2.19) Human Scale Score = 2.61 + (0.4*-0.74) + (0.238*1.10) + (h33*-0.003) + (h34*0.05) + (h35*0.04) Transparency Score = 1.71 + (t41*1.22) + (t42*0.67) + (t43*0.53) Complexity Score = 2.61 + (c51*0.05) + (c521*0.23) + (c522*0.12) + (c53*0.42) + (c54*0.29) + (c55*0.03) Where i17 and c53 are dummy variables: i17 = 0 (if outdoor dining is absent); and i17 = 1 (if outdoor dining is present). c53 = 0 ( if outdoor dining is absent); and c53 = 1 (if outdoor dining is present). 3.2. Control Variables Property characteristics were used at the property level, and built environment characteristics were used at the street segment (neighborhood) level as control variables. GIS data for the study area, including MapPluto parcel layers and street segment centerlines, were acquired from the New York City Department of City Planning. MapPluto files contain each lot's tax data, with the lot's physical features stored in ESRI shape file and database table file formats. These data categorized the buildings by use/type, such as office, retail, or residential. At the property level, the authors included building age (BUIDLING AGE), building lot frontage facing the street (LN LOT FRONT), and Floor-Area-Ratio (FAR), capturing the physical characteristics of the properties. The authors also controlled for single-family residential properties using a dummy variable (SINGLE RES). At the street segment level, specific variables were included to represent accessibility to destinations likely to be related to property values. First is the degree of walkability captured by Walk Score®, a web-based platform that collects the number of amenities (e.g., grocery stores, restaurants, schools) within a walkable distance of a specific address in order to rate its walkability on a numeric scale ranging from 0 to 100. Drawing on Walk Score® data, this study used the walk score for each street segment using the physical address of properties in the center of each block face (from an intersection to the next intersection) (WALK SCORE). The next variable included in the model is distance to transit (NET DISTANCE TRANS). Proximity to transit is heavily studied in hedonic models as a significant determinant of property value. Using ESRI ArcInfo Network Analyst and New York City road centerline shape files, the authors measured the shortest distance (in miles) to the closest rail station from each study segment center point. Additionally, the distance to CBD (DISTANCE CBD), also in miles, were measured, since this could be associated with property values. Mixed land use is often considered contributor to pedestrian-friendly development, potentially correlated with property values. Entropy (BLOCK ENTROPY) was used as a measure of mixed land use, in the census block where the property is located, using Formula 2, borrowed from Ewing and Clemente (2013) . Formula 2: entropy = −[residential share*LN (residential share) + retail share*LN (retail share) + office share*LN (office share)]/LN (3). … where the shares were computed based on floor area of each use for lots within the buffer (entropy). 3.3. Analytical Methods The independent variables in this study come from two geographic levels: property level and street level. Since all properties in the same block share the same urban design qualities and the built environmental characteristics (e.g., imageability, distance to rail) of the street segment in which properties are located, a nesting structure tends to emerge. This nesting structure generates correlation between observations, and violates the assumption of independence of errors attributed to Ordinary Least Squares (OLS). In this case, OLS may not accurately reveal the variances and may underestimate the standard errors of regression coefficients. The appropriate statistical method for nested data is Multilevel Modeling (MLM), also known as Hierarchical Modeling (HLM) ( Goldstein, 2011 ; Heck & Thomas, 2015 ). It allows for overcoming the limitations of OLS and controlling for dependence among lower level cases, and hence yields more precise coefficients and standard error estimations. The authors employed HLM 6.08 software to perform the MLM and to control for the dependence of properties within a given block on the street segment characteristics (Formula 3). Formula 3: Level-1 Model. f ( Ln PROPERTY VALUE ) = B 0 + B 1 *(Ln LOT FRONT) + B 2 *(Ln BUILDING AGE) + B 3 *(FAR) + B 4 *(SINGLE RES). Where SINGLE RES is a dummy variable: SINGLE RES = 0 (if the unit is not a single-family). SINGLE RES = 1 (if the unit is a single-family). Level-2 Model. B 0 = G 00 + G 01 *(IMAGEABILITY) + G 02 *(ENCLOSURE) + G 03 *(HUMAN SCALE) + G 04 *(TRANSPARENCY) + G 05 *(COMPLEXITY) + G 06 *(DISTANCE CBD) + G 07 *(BLOCK ENTROPY) + G 08 *(WALK SCORE) + G 09 *(NET DISTANCE TRANS). Initially, the authors ran a correlations matrix to see the potential association between the independent variables in both levels and property values. At level-1, FAR, lot front area are positively associated with property values (Pearson Correlation (PC) 0.611 and 0.615 respectively) and building age is negatively associated with property value with PC value of −0.128. Also, the correlation matrix demonstrates that, at level-2, the distance to transit and also to CBD exhibit negative relationship (PC of −0.224 and − 0.424, respectively) while Walkscore shows a positive association with property values (PC of 0.349). Finally, the authors tested for multi-collinearity using the VIF threshold of 10 following by (Hair et al. 2009). 2 Publisher Copyright: © 2019 Elsevier Ltd

PY - 2020/3

Y1 - 2020/3

N2 - Urban design qualities have the potential to contribute to the sense of safety, comfort, engagement, and overall neighborhood satisfaction perceived by residents, thus can be related to higher property values. Yet, many of these urban design qualities are highly conceptual, require extensive data collection and are subject to various interpretations. As a result, there is little empirical evidence on how street-level urban design qualities are related to property values. Drawing on Multi-level Modeling, this article employs a highly cited dataset on urban design qualities in New York City to provide a statistical analysis of the direct relationship between these qualities and property values. Controlling for confounding factors, this article identifies “imageability” in the street-level environment as a featured urban design quality with the most statistically significant association with property values. In addition, research findings suggest that “transparency” of building facades are positive predictors of property values, whereas the complexity of the built environment exhibits a negative correlation with property values. Policy implications for planners, urban designers, and developers include designing guidelines and street layouts that encourage memorable civic image and identity. While investing in building facades with greater transparency would yield higher property values, complexity in the urban environment, particularly in neighborhoods seeking investment and attracting capital should be treated with care by all parties involved.

AB - Urban design qualities have the potential to contribute to the sense of safety, comfort, engagement, and overall neighborhood satisfaction perceived by residents, thus can be related to higher property values. Yet, many of these urban design qualities are highly conceptual, require extensive data collection and are subject to various interpretations. As a result, there is little empirical evidence on how street-level urban design qualities are related to property values. Drawing on Multi-level Modeling, this article employs a highly cited dataset on urban design qualities in New York City to provide a statistical analysis of the direct relationship between these qualities and property values. Controlling for confounding factors, this article identifies “imageability” in the street-level environment as a featured urban design quality with the most statistically significant association with property values. In addition, research findings suggest that “transparency” of building facades are positive predictors of property values, whereas the complexity of the built environment exhibits a negative correlation with property values. Policy implications for planners, urban designers, and developers include designing guidelines and street layouts that encourage memorable civic image and identity. While investing in building facades with greater transparency would yield higher property values, complexity in the urban environment, particularly in neighborhoods seeking investment and attracting capital should be treated with care by all parties involved.

KW - Complexity

KW - Enclosure

KW - Imageability

KW - Property values

KW - Residential design premiums

KW - Transparency

KW - Urban design

UR - http://www.scopus.com/inward/record.url?scp=85077316209&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85077316209&partnerID=8YFLogxK

U2 - 10.1016/j.cities.2019.102564

DO - 10.1016/j.cities.2019.102564

M3 - Article

AN - SCOPUS:85077316209

SN - 0264-2751

VL - 98

JO - Cities

JF - Cities

M1 - 102564

ER -

Do Urban Design qualities add to property values? An empirical analysis of the relationship between Urban Design qualities and property values

Abstract

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this