12
100 Valuing New Development in Distressed Urban Neighborhoods Does Design Matter? Brent D. Ryan and Rachel Weber A housing construction boom occurred in some of the poorest urban neigh- borhoods in the United States in the 1990s. Attracted by vacant land and new markets, and possessing access to cheap credit, for-profit developers built a mix of housing, ranging from multifamily buildings to gated single-family homes in poor neighborhoods. The urban design of this new housing varied widely. In this article, we examine whether urban design is a significant contributor to the value of new housing in poor urban neighborhoods, assuming that resident preferences are revealed in the prices paid for different kinds of housing and that these in turn are reflected in their assessed values. We distinguish between three urban design types: enclave, traditional neighborhood development (TND), and infill. We perform a parcel-based hedonic regression to explain the values of new housing constructed in high-poverty Chicago census tracts between 1993 and 2003. We investigate the relationship between urban design and housing values in poor neighborhoods, about which little is known, because previous research on the effects of urban design on housing values has focused almost exclusively on new urbanist projects in more affluent areas. We also hope to make local governments aware of the potential of urban design policies to create value in distressed neighborhoods and to reduce resistance to new development products among realtors and tax assessors who shape real estate market practices. The Urban Design of New Inner-City Housing Substantial amounts of privately financed housing have been constructed in distressed, inner-city neighborhoods during the past decade (Ryan, 2006a), transforming them through an influx of higher-income residents (Jargowsky, 2003; Wyly & Hammel, 1999) and capital. Urban design is particularly important in this context. In poor areas with large amounts of vacant land, developers can sometimes acquire whole city blocks and reshape street networks (Ryan, 2006b), creating more design options than elsewhere. Urban design may also reduce the social isolation of low-income households, enhancing their integration into the larger urban economy (Duany, Plater-Zyberk, & Speck, 2000; U. S. Department of Housing and Urban Development [HUD] 2000; Wilson, 1996). We estimate the effect of design on the assessed values of new housing units in high-poverty Chicago census tracts with a parcel-based hedonic regression in which we distinguish between three urban design types: enclave, traditional neighborhood development (TND), and infill. We find that urban design signifi- cantly affects housing values, and infill housing is more highly valued than either enclave or TND housing. We also examine the influences of individual urban design features and find that residents prefer entrances that face the street, and facades constructed from the same material as adjacent buildings. They also prefer parking in front of their homes, and to be buffered from public streets. We interpret the former to be preferences for greater integration into the surrounding neighbourhood, consistent with our findings on infill. Brent D. Ryan, AICP ([email protected]), is co-director of the City Design Center and an assistant professor of urban planning and policy at the University of Illinois at Chicago. His research interests include urban design, neighborhood revitaliza- tion, and morphological change in urban areas. Rachel Weber ([email protected]) is an associate professor of urban planning and policy at the University of Illinois at Chicago. She is the author of numerous articles, technical reports, and a book in the fields of development finance, urban real estate markets, and industrial restructuring. Journal of the American Planning Association, Vol. 73, No. 1, Winter 2007 © American Planning Association, Chicago, IL. JAPA 73-1 07 Ryan fin 1/15/07 6:06 PM Page 100

Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

100

Valuing New Developmentin Distressed UrbanNeighborhoods

Does Design Matter?

Brent D. Ryan and Rachel Weber

Ahousing construction boom occurred in some of the poorest urban neigh-borhoods in the United States in the 1990s. Attracted by vacant land andnew markets, and possessing access to cheap credit, for-profit developers

built a mix of housing, ranging from multifamily buildings to gated single-familyhomes in poor neighborhoods. The urban design of this new housing variedwidely.

In this article, we examine whether urban design is a significant contributorto the value of new housing in poor urban neighborhoods, assuming that residentpreferences are revealed in the prices paid for different kinds of housing and thatthese in turn are reflected in their assessed values. We distinguish between threeurban design types: enclave, traditional neighborhood development (TND), andinfill. We perform a parcel-based hedonic regression to explain the values of newhousing constructed in high-poverty Chicago census tracts between 1993 and2003. We investigate the relationship between urban design and housing valuesin poor neighborhoods, about which little is known, because previous researchon the effects of urban design on housing values has focused almost exclusivelyon new urbanist projects in more affluent areas. We also hope to make localgovernments aware of the potential of urban design policies to create value indistressed neighborhoods and to reduce resistance to new development productsamong realtors and tax assessors who shape real estate market practices.

The Urban Design of New Inner-City Housing

Substantial amounts of privately financed housing have been constructedin distressed, inner-city neighborhoods during the past decade (Ryan, 2006a),transforming them through an influx of higher-income residents (Jargowsky, 2003;Wyly & Hammel, 1999) and capital. Urban design is particularly important inthis context. In poor areas with large amounts of vacant land, developers cansometimes acquire whole city blocks and reshape street networks (Ryan, 2006b),creating more design options than elsewhere. Urban design may also reduce thesocial isolation of low-income households, enhancing their integration into thelarger urban economy (Duany, Plater-Zyberk, & Speck, 2000; U. S. Departmentof Housing and Urban Development [HUD] 2000; Wilson, 1996).

We estimate the effect of design on theassessed values of new housing units inhigh-poverty Chicago census tracts witha parcel-based hedonic regression inwhich we distinguish between threeurban design types: enclave, traditionalneighborhood development (TND), andinfill. We find that urban design signifi-cantly affects housing values, and infillhousing is more highly valued thaneither enclave or TND housing. We alsoexamine the influences of individual urbandesign features and find that residentsprefer entrances that face the street, andfacades constructed from the samematerial as adjacent buildings. They alsoprefer parking in front of their homes, andto be buffered from public streets. Weinterpret the former to be preferences forgreater integration into the surroundingneighbourhood, consistent with ourfindings on infill.

Brent D. Ryan, AICP ([email protected]), isco-director of the City Design Center andan assistant professor of urban planningand policy at the University of Illinois atChicago. His research interests includeurban design, neighborhood revitaliza-tion, and morphological change in urbanareas. Rachel Weber ([email protected])is an associate professor of urban planningand policy at the University of Illinois atChicago. She is the author of numerousarticles, technical reports, and a book inthe fields of development finance, urbanreal estate markets, and industrialrestructuring.

Journal of the American Planning Association,

Vol. 73, No. 1, Winter 2007

© American Planning Association, Chicago, IL.

JAPA 73-1 07 Ryan fin 1/15/07 6:06 PM Page 100

Page 2: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

Ryan and Weber: Valuing New Development in Distressed Urban Neighborhoods 101

Studies of distressed neighborhoods in Detroit andPhiladelphia (Ryan, 2002, 2006b), Pittsburgh (Dietrick &Ellis, 2004), and our preliminary observations in Chicago,show that new housing development in poor neighborhoodscan be grouped into three urban design types: (1) infill, orscattered-site, development; (2) traditional neighborhooddevelopment (TND); and (3) enclave, or self-contained, de-velopment. Table 1 details some characteristics of each type.

Infill development (illustrated in Figure 1) occurswhere small numbers of parcels are available for redevelop-ment on existing city blocks. This type of development doesnot change the neighborhood structure substantially be-cause new housing is located between existing buildingsoriented to current street and lot subdivision patterns.TND and enclave developments (examples in Figures 2and 3) occur where empty parcels are numerous enoughto permit the construction of extensive, contiguous, newhousing. These latter types allow designers much moreflexibility in how they locate housing, open space, roadways,and parking areas.

TNDs integrate new development into their surround-ings by replicating the design features of existing neighbor-hoods, like street-facing housing and interconnected streetgrids. TNDs have much in common with new urbanistdesigns (Bothwell, Gindroz, & Lang, 1998; Morrow-Jones,Irwin, & Roe, 2004; Talen, 2001; Steuteville, 1999; Leccese& McCormick, 2000). In contrast, enclaves reject theircontexts by spatially isolating new housing from their sur-roundings through the orientation and spatial placement ofbuildings and roadways. Bohl (2000) refers to enclave de-velopments as “inward-focused residential pods” (p. 767).

Urban Design’s Effect on Housing Value

Urban economists have demonstrated that a property’sattributes affect its price in ways that can be measured (seeBoyle & Kiel, 2001; Sirmans, MacPherson, & Zietz, 2005for literature reviews). Few economists have specifically con-

sidered the design of the built environment (or if they haveit has been in the context of new suburban developments)even though differences in urban design might affecthousing prices by influencing development costs, amenities,and uncertainty about future development nearby.

Development CostsEach of the three types of urban design described

above uses space in a different manner, potentially affectingconstruction cost, sale price, and assessed value. Housingunits in enclaves and TNDs are likely to cost less to buildthan comparable infill units because they can take advan-tage of economies of scale (Gyourko & Rybczynski, 2001),and may also have access to cheaper capital. Enclaves andTNDs are also likely to have lower per-unit legal and othercosts of buying property compared to infill development.In other ways, enclaves and TNDs may be more expensiveto build. They may require more land per unit than infilldevelopment, for example, for roadways, landscaping, andparking areas.

Amenities and DisamenitiesOur three urban design types also produce different

types of amenities. Enclaves and TNDs often provideadditional site design amenities such as new infrastructure,landscaping, and convenient off-street parking. Parking forinfill development by contrast, may be less safe because ofheavier traffic and crime on alleys and streets.1 The differentdesign types also face differing constraints that influencetheir locational amenities. For example, enclaves and TNDsrequire large parcels, limiting their feasible locations morethan is the case for infill. Finally, differences among designtypes may influence social interactions, which in turn in-fluence residents’ safety and participation in neighborhoodcivic life (Newman, 1972; Bothwell et al., 1998; Duany etal., 2000; Jacobs, 1961; Whyte, 1988). Scholars haveargued that existing urban neighborhoods and TNDsencourage more social interaction, and some have eventried to quantify these attributes. For example, Eppli and

Table 1. General characteristics of different urban design types.

Number of Number ofdevelopers urban design Physical

per unit decisions integrationDevelopment type Parcel size of land area possible with context

Infill Small Many Few HighTraditional neighborhood development (TND) Large Few Many ModerateEnclave Large Few Many Low

JAPA 73-1 07 Ryan fin 1/15/07 6:06 PM Page 101

Page 3: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

Tu (1999) found that new-urbanist-style developmentscommanded higher prices than similar, suburban-styleunits. Song and Knaap (2003) found that residents ofnew urbanist communities were willing to pay more fordevelopment designed for internal connectivity, but not fordevelopment designed to be integrated with the surroundingenvironment.

New housing that is poorly integrated with its sur-roundings may stigmatize residents, particularly if they arelow-income. Recent redevelopments of public housingprojects have replaced enclave design features with TNDfeatures to integrate this housing better into its surround-ings (HUD & Congress for the New Urbanism, 2000).Infill housing is not spatially isolated, and does not distin-guish itself from its context as enclaves and TNDs do. Thisgives residents of new infill less ability to control access byoutsiders, and thus risk, than residents of enclaves and

TNDs have, but this problem may be offset by improvedcommunity social controls, particularly if residents knoweach other well (Bothwell et al., 1998; Song & Knaap,2003).

Future UncertaintyIt is a disadvantage when potential homebuyers “do

not know with certainty how the neighborhood develop-ment will evolve or proceed over time” (Sirmans, Turnbull,& Dombrow, 1997, p. 615; see also Thorsnes, 2000).Different urban design types are associated with differentlevels of future uncertainty. The homogeneous nature ofan enclave or TND assures purchasers that future units willbe similar to existing ones.2 This is generally less true ofinfill. One can therefore argue that enclave and TND modelshelp to internalize some of these potential negative externali-ties, and residents may be willing to pay a premium for this.

102 Journal of the American Planning Association, Winter 2007, Vol. 73, No. 1

Figure 1. Typical infill-type housing.

JAPA 73-1 07 Ryan fin 1/15/07 6:06 PM Page 102

Page 4: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

Ryan and Weber: Valuing New Development in Distressed Urban Neighborhoods 103

Because the influences described above work againstone another, and their magnitudes are unknown, theliterature does not permit strong a priori hypotheses aboutwhich urban design type (infill, TND, or enclave) will bethe most desired and therefore most highly valued. Thefollowing sections describe our empirical investigation toreveal preferences for the urban design of new housing inlow-income neighborhoods.

Data Collection and Analysis

We assembled construction permit data on all parcels(land lots and their built improvements, if any) on whichhousing units were constructed between January 1, 1993and December 31, 2001 in census tracts where at least20% of households had incomes below the federal poverty

line in 1990. This number is widely accepted as a thresh-old for neighborhood distress (Galster, 2002; Jargowsky1997). Forty-six percent of Chicago census tracts weredistressed in 1990 using this measure. We excluded fromour analysis census tracts within 2 miles of the CentralBusiness District (CBD). Although there were distressedneighborhoods inside this perimeter in 1990, infrastruc-ture improvements and massive amounts of new publicinvestment since that time made them highly unusual.

We associated Cook County Assessor’s Office data onnew condominiums, attached and detached single-familyhomes, and apartment buildings with six or fewer units,3

with construction permit addresses. Eighty-six percent ofthe building permit addresses could be matched to ParcelIdentification Numbers (PINs)4, yielding 1,227 parcels forwhich we had both construction permit addresses andassessment data. Within this group of records, we had

Figure 2. Typical traditional neighborhood development-type housing.

JAPA 73-1 07 Ryan fin 1/15/07 6:06 PM Page 103

Page 5: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

complete information for a subset of 823 parcels, includingcritical data on the building characteristics of the structureson each parcel. We report results for both larger (N = 1,227)and smaller (n = 823) samples for variables for which wehad data in each case.

Dependent VariableWe sought to explain housing value, which we measured

using 2003 assessed values. Because we had parcel-leveldata, our dependent variable was the assessed value of anentire parcel, not of an individual dwelling unit.5 In CookCounty, parcels are supposed to be assessed at 16% of theirestimated market values. We relied on assessments insteadof housing unit sales prices for several reasons. First, inlow-income neighborhoods, where home ownership is lesscommon, only 17% of our small sample could be matchedto sales transaction data.6 Second, examining only sold

properties may introduce selection bias if this sample issignificantly different from the unsold ones (Gatzlaff &Haurin, 1997). Third, assessments are good proxies formarket values, as each Chicago parcel is reassessed everythree years based on any recent sales of the parcel in ques-tion and on sales of comparable parcels.7 Although the useof assessed values may introduce some degree of error intothe model, we felt it was likely to be randomly distributed.

The Independent Variable of GreatestInterest: Urban Design Type

We hypothesized that urban design would have aneffect on assessed housing values even after other attributesthat might influence demand were controlled. Determiningthe urban design type of an individual parcel required us todefine the “cluster” of similar new units to which it be-longed. Since TND and enclave developments contain

104 Journal of the American Planning Association, Winter 2007, Vol. 73, No. 1

Figure 3. Typical enclave-type housing.

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 104

Page 6: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

Ryan and Weber: Valuing New Development in Distressed Urban Neighborhoods 105

multiple units by definition, we needed groups of infillunits which would be comparable to these.8

We first address-matched all qualifying building permitsto a GIS database, defined a 250-foot buffer around theaddress on each permit, and joined overlapping buffers tocreate clusters. We felt that units separated by more than250 feet would not be visible or closely accessible to eachother, reducing the appropriateness of defining them as agroup. We then eliminated clusters with fewer than 20total units, since smaller developments might not have theurban design features needed to identify TND- and enclave-type developments.

We visited and photographed each cluster and matchedthese field data to high-resolution aerial photographs andGIS figure-ground illustrations to confirm infill clusters(Google, 2005; City of Chicago Department of Planningand Development, 2005; Cook County Office of theAssessor, 2005), then classified each remaining (non-infill)PIN in our sample as either an enclave or TND by deter-mining whether the following were present or absent:

1. parking (either a lot or individual spaces) in front;2. roadways interior to the lot, such as driveways or

access roads;3. front doors opening onto interior walkways,

roadways, or private open space;4. extensive buffering (substantial trees, plantings,

open space, or landscaped berms) between thebuilding and the street; and

5. façade materials which differed from those ofadjoining buildings.

Developments possessing three or more of the foregoingattributes we considered to be enclaves, and those lackingthree or more we considered to be TNDs. We used thesecriteria as binary variables in later model specifications.Using multiple design criteria also permitted us to catego-rize developments that possessed only some of the designfeatures and to analyze developments with mixed designfeatures.

Figure 4 shows the geographic distribution of oursample, and Table 2 shows how the sample broke down bydesign type. The distribution was similar in both samples.In both samples average assessed values were significantlyhigher for infill clusters than for either enclave or TNDclusters.

Other Independent VariablesIn addition to dummy variables for urban design types

and parcel attributes, we also included other site-specificvariables likely to influence parcel value. Descriptive statis-

tics for these and the urban design variables are shown forboth the small and large samples in Table 3. 9

Model

We employed a standard hedonic model to regressurban design features on the assessed values of parcels withnew construction in high-poverty Chicago census tracts.We adopted the following semi-log functional form becauseof an observed nonlinear relationship between assessedvalue and key parcel attributes like lot size (see Colwell andMunneke, 1997):

Ln(Assessed value) = a + bX + dZ + gSCALE + lENCLAVE + wTND + e

In the equation above, the dependent variable is thenatural log of a parcel’s assessed value in 2003, a is theintercept, X represents a vector of characteristics of thestructure on the parcel, Z a vector of neighborhood attri-butes, SCALE represents number of units in the cluster,and e represents an error term. The binary variables ofENCLAVE and TND each take on values of 1 if the parcelis located in an enclave or TND cluster and 0 otherwise.These two dummy variables are mutually exclusive.

Results

Table 4 shows our regression results. The adjusted R2

values range from 28% to 86%, indicating that the ex-planatory power of the models is, in some cases, very high.In most cases, the coefficients on the independent variablesare as expected. Homes wiath the following attributes wereassessed at higher values: more bedrooms and bathrooms;recent sales; and locations in higher-income areas, near theCBD, near Lake Michigan, and near transit stops. However,we are primarily interested in the urban design variables.Using both the small and large samples, the coefficients onthe dummy variables for location in an enclave or TNDare negative in Models 1 and 3. These results suggestlocating in these types of development reduces value: anidentical building constructed as infill is worth muchmore. Specifically, location in an enclave decreases housingvalue by between 22% (large sample) and 24% (smallsample), and location in a TND development decreasesvalue by between 21% (small sample) and 27% (largesample) compared to the same unit built as infill.10

We then sought to discover which of the individualdesign elements characteristic of TNDs and enclaves

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 105

Page 7: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

106 Journal of the American Planning Association, Winter 2007, Vol. 73, No. 1

Figure 4. Location and urban design type of sample construction permits.

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 106

Page 8: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

housing consumers apparently do not like. In Models 2and 4 we compared only TND and enclave parcels (i.e. weexcluded infill). We found the majority of the coefficientson individual design element dummy variables to besignificant under these conditions, and we were able toincrease the explanatory power of the original models bysubstituting these criteria for the more general variablesrepresenting urban design types.

Specifically, our results showed residents to prefersome buffer between their living quarters and the street.They also preferred to have parking adjacent to the street,in front of their homes. These urban design features arecharacteristic of enclaves, and serve to separate housingfrom its surroundings. Two other variables, building mate-rial different from adjoining, and opens to the yard, hadnegative and significant coefficients, suggesting that resi-dents prefer to be more integrated into their surroundings.Street-facing building entrances and contextual façades,both typical of TNDs, increase the value of properties incontiguous developments. The fifth individual variable, thepresence or absence of a private road, was contradictory:for the smaller sample it was an asset, while for the largersample it was a liability.

Conclusions

Our findings indicate that urban design plays a mean-ingful role in determining housing values in low-incomeChicago neighborhoods. Most importantly, infill housingappears to command a value premium, compared to bothTND and enclaves. From this we understand consumers tovalue housing that is integrated into its urban context over

housing which is dissociated from it. People may associateurban developments that are homogeneous and dissociatedfrom their surroundings with public housing, particularlyin low-income neighborhoods.

We also conclude that the value penalty associatedwith TND and enclave developments could be reduced bybetter connecting these developments to the existing urbanfabric. Two individual characteristics (front parking andstreet buffering) had positive impacts, while two others(private roadways and non street-facing entrances) hadnegative impacts. We conclude that residents valued indi-vidual urban design elements of both the enclave and TNDmodels. They seemed to appreciate the convenience andsafety of accessible, visible, parking in front of units; theprivacy provided by separation from the street; and beinga part of their surroundings, as expressed by contextualfacades, street-facing entrances, and a shared public street.Our findings are consistent with those of previous re-searchers who found similar preferences among suburbandwellers (Morrow-Jones et al., 2004; Song & Knaap, 2003;Talen, 2001).

One caveat is that some of the observed value differ-ential may be due to different land acquisition and devel-opment costs for the different design types. However,interviews with housing developers active in these neigh-borhoods suggest that lack of economies of scale may addto the cost of infill development, and that enclaves requiremore roadways and landscaping than infill.11 Moreover,the design criteria variables remained statistically significantin the models run on data that did not include parcelsdeveloped as infill (Models 2 and 4).

We found that whether contiguous developments aredesigned as enclaves or TNDs they are less valuable than

Ryan and Weber: Valuing New Development in Distressed Urban Neighborhoods 107

Table 2. Percent of sample devoted to each of three urban design types and mean assessed values.

Percent of samples Mean assessed value(SD)

Small sample Large sampleDevelopment type (n = 823) (N = 1,227) Small sample Large sample

Enclave 41% 36% $32,909 $31,872(16,224) (14,723)

TND 21% 20% $31,372 $30,159(18,466) (16,467)

Infill 38% 44% $49,688 $41,573(18,220) (18,083)

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 107

Page 9: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

infill housing. This confirms the work of those theorists,beginning with Jane Jacobs, who have argued that urbandevelopment that is integrated is more desirable than thatwhich is isolated. Our results, showing that both theenclave- and TND-style design models carry similar valuepenalties, challenge the neotraditionalist argument thatTNDs are superior to other models of urban design (seeDuany et al., 2000).

Our results should reassure those who believe that thebest way to revitalize urban neighborhoods is to respectand augment the urban design character of existing placesrather than to transform them in more dramatic ways.Cities may want to consider these findings as they establishboth redevelopment guidelines and formal and informaldesign standards for publicly assisted housing in distressedneighborhoods.

108 Journal of the American Planning Association, Winter 2007, Vol. 73, No. 1

Table 3. Attributes of sampled parcels with new construction and the distressed Chicago census tracts where they are located, 2003.

Small sample (n = 823) Large sample (N = 1,227)

Min Max Mean SD Min Max Mean SD

Assessed value 2,946 126,564 38,943 19,390 2,057 126,564 35,750 17,372

Full baths 1 10 2.27 1.26

Bedrooms 2 18 3.5 1.9

Masonry exterior construction? 0 1 .81 .39(0 = No, 1 = Yes)

Square feet of land 520 34,000 1,957 1,680

Units in parcel 1 6 1 .82 1 7 1.27 .87

Units in cluster 6 240 82 83 6 240 83 85

Age of unit (years) 1 10 6.07 2.07 1 10 5.47 2

Tract median household income ($) 12,599 95,075 46,511 21,140 12,599 95,075 49,182 22,041

Tract percent owner-occupied 4% 71% 39% 14% 4% 71% 39% 16%

Tract percent Black 1% 98% 38% 42% 1% 98% 31% 39%

Tract percent Hispanic 0% 80% 22% 21% 0% 80% 23% 20%

Any recent sale? (0 = No, 1 = Yes) 0 1 .17 .38 0 1 .12 .32

Distance to CBD (miles) 2.01 7.67 4.02 1.30 2.00 9.69 3.91 1.34

Distance to Lake Michigan (miles) .22 5.06 2.29 1.30 .06 5.06 2.27 1.16

Distance to elevated rail stop (miles) .06 1.31 .55 .30 .06 1.31 .52 .28

Percent change in quartersection’s equalized assessed value, 1989–1997 46% 588% 287% 149% 21% 588% 317% 151%

Percent of quartersection’s equalized assessed value in commercial and industrial uses 5% 70% 37% 16% 5% 70% 36% 15%

In an enclave? (0 = No, 1 = Yes) 0 1 .41 .49 0 1 .36 .48

In a TND? (0 = No, 1 = Yes) 0 1 .21 .41 0 1 .20 .40

Infill? (0 = No, 1 = Yes) 0 1 .38 .49 0 1 .44 .50

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 108

Page 10: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

Ryan and Weber: Valuing New Development in Distressed Urban Neighborhoods 109

Table 4. Results of regression model predicting 2003 assessed value for sampled parcels.

Small sample Large sample

Model 1: Model 2: Design Model 3: Model 4: DesignDesign type characteristics Design type characteristics

(t) (t) (t) (t)

Full baths 0.029 0.046*(1.928) (2.234)

Bedrooms 0.078** 0.089**(6.125) (5.994)

Masonry exterior construction 0.029 0.025(0 = No, 1 = Yes) (1.003) (0.886)

Square feet of land −.000 0.000 −0.000** −0.000**(−0.073) (1.705) (−5.215) (−9.781)

Units in parcel −0.112** −0.116** 0.048** −0.019(−3.789) (−2.974) (3.195) (−1.135)

Units in cluster −0.000 0.010** 0.001 0.000(−1.511) (9.213) (1.479) (0.723)

Age of unit (years) −0.029** 0.004 0.019** −0.013(−4.973) (0.687) (2.717) (−1.419)

Tract median household income 0.000** 0.000** 0.000** 0.000**(11.673) (5.410) (6.934) (8.946)

Tract percent owner-occupied −1.679** −3.402** −0.847** −1.710**(−9.069) (−5.848) (−4.480) (−3.772)

Tract percent Black −0.389** 8.894** −0.201* 1.945**(−6.010) (10.834) (−2.418) (4.703)

Tract percent Hispanic −0.365** 20.919** −0.005 3.320**(−3.341) (11.543) (−0.038) (4.130)

Any recent sale? (0 = No, 1 = Yes) 0.077** 0.038(2.696) (1.417)

Distance to CBD (miles) −0.110** −0.563** −0.040** −0.104**(−7.774) (−6.492) (−3.176) (−5.604)

Distance to Lake Michigan (miles) −0.406** −0.491** −0.312** −0.099(−1.574) (−3.057) (−4.694) (−0.563)

Distance to closest elevated rail stop −0.020 −0.173** −0.060** −0.370**(miles) (−7.495) (−3.446) (−3.733) (−8.306)

Percent change in quartersection’sequalized assessed value, 1989–1997 −0.001** 0.019** −0.000 0.000**

(−7.041) (11.381) (−1.370) (3.216)

Percent of quartersection’s equalized assessed value in commercial and industrial uses −0.543** −5.024** −0.139 0.496*

(−6.164) (−12.192) (−1.304) (2.536)

In an enclave? (0 = No, 1 = Yes) −0.219** −0.203**(−4.765) (−3.496)

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 109

Page 11: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

AcknowledgementsThis research was funded by a Lincoln Institute of Land Policy Planningand Development Fellowship. We greatly appreciate the researchassistance we received from Dan Weiske and Nina Savar and feedbackfrom participants in the Lincoln Institute of Land Policy Planning andDevelopment seminar.

Notes1. On-street parking may reduce the appeal of nearby developments bycongesting streets and reducing the “aesthetic appeal of the neighbor-hood” (Guttery, 2002, p. 266; see also Bohl, 2000). Other scholarsdisagree, citing narrow streets with on-street parking and slower trafficas a positive contributor to perceptions of resident comfort and safety(Appleyard, Lynch, & Mier, 1966).2. Prior research has found that contiguous developments lend them-selves to institutional arrangements, such as restrictive covenants, thatreduce future risks of negative neighborhood effects (see Alexandrakis &Berry, 1994; Hughes & Turnbull, 1996; Speyrer 1989). Peiser (1984)found slightly higher net benefits to “planned” (i.e., large-scale) versusunplanned developments and Ellen, Schill, Susin, and Schwartz (2001)found that larger-scale and denser developments had significantly largereffects on values in the surrounding areas.

3. We largely avoided issues raised by subsidized housing developmentsby excluding from our sample Class 4 parcels, which are those developedby nonprofits.4. We assumed that construction permits whose addresses we could notmatch to assessment data either had incorrect address information, werenot built in time to be assessed in 2003, or had been built on newlysubdivided parcels. In order to determine if we were introducing biasinto the sample by requiring a match, we regressed critical locationaldata (e.g., distance to CBD) against a binary variable indicating whetherthe construction data were matched or unmatched. In none of theseregressions was this variable ever statistically significant, and so weconcluded that successful matches were spatially random.5. We do, however, account for the number of units in each parcel byincluding this information as an independent variable.6. We expect that the assessor has access to more complete transactionsdata and that, in reality, a larger share of our sample did indeed sell.7. When a new building is built, the assessor reviews construction costinformation from the building permit and acquires any sales data. Themost recent sale price is a baseline market value that will be checkedagainst adjustment factors generated by regressions of area sale prices.8. We developed the following technique to avoid over- and under-sampling from clusters. We divided the “total construction value” listedon the permit for the proposed project by assumed construction costs of

110 Journal of the American Planning Association, Winter 2007, Vol. 73, No. 1

Table 4 (continued).

Small sample Large sample

Model 1: Model 2: Design Model 3: Model 4: DesignDesign type characteristics Design type characteristics

(t) (t) (t) (t)

In a TND? (0 = No, 1 = Yes) −0.195** −0.243**(−4.291) (−5.273)

Building material different fromadjoining? (0 = No, 1 = Yes) −1.840** −0.412**

(−11.084) (−3.283)

Served by private road? (0 = No, 1 = Yes) 1.545** −0.591**(8.199) (−3.343)

Parking lot in front? (0 = No, 1 = Yes) 3.853** 1.130**(13.031) (5.372)

Opens to yard? (0 = No, 1 = Yes) −5.227** −0.627**(−12.238) (−2.833)

Buffered from the street? 1.513** 0.914**(0 = No, 1 = Yes) (7.132) (5.102)

Constant 11.635** −2.638* 10.661** 8.873**(78.419) (−2.398) (71.190) (23.529)

Adjusted R 2 .679 .862 .282 .496

N 823 511 1227 692

*p < 0.05 **p < 0.01

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 110

Page 12: Valuing New Development in Distressed Urban Neighborhoodsdusp.mit.edu/sites/dusp.mit.edu/files/attachments/...distressed neighborhoods and to reduce resistance to new development products

$75,000 per unit, to obtain the number of units. If we could matchfewer than 20% of these expected units with their PINs from theAssessor’s Office, we eliminated the cluster to avoid under-sampling it.In clusters where we matched over 75% of the expected units, werandomly eliminated PINs to reduce the match rate to no more than75% to avoid over-sampling.9. Ideally, we would have also controlled for the housing tenure of eachparcel as well as its land and development costs. Unfortunately, suchdata are considered proprietary information and is generally unavailable.10. In a semi-log regression, the coefficient on a dummy variable can beinterpreted as an elasticity as follows: when TND and ENCLAVEchange from 0 to 1, the value of the parcel will change by [Exp(b) −1] ×100%.11. We conducted interviews with Thrush Development Corporationand Applied Real Estate Analysis (AREA).

ReferencesAlexandrakis, E., & Berry, B. (1994). Housing prices in master plannedcommunities: Are there premiums? The evidence for Collin County,Texas 1980–1991. Urban Geography, 15 (1), 9–24.Appleyard, D., Lynch, K., & Mier, J. R. (1966). The view from the road.Cambridge, MA: MIT Press.Bohl, C. (2000). New urbanism and the city: Potential applications andimplications for distressed inner city neighborhoods. Housing PolicyDebate, 11 (4), 761–820.Bothwell, S., Gindroz, R., & Lang, R. (1998). Restoring communitythrough traditional neighborhood design: A case study of Diggs TownPublic Housing. Housing Policy Debate, 9 (1), 89–112.Boyle, M., & Kiel, K. (2001). A survey of house price hedonic studiesof the impact of environmental externalities. Journal of Real EstateLiterature, 9 (2), 117–144.City of Chicago, Department of Planning and Development. (2005).Maps of Chicago. Retrieved October 15, 2005, from http://maps.cityofchicago.org/website/public/viewer.htmColwell, P. F., & Munneke, H. J. (1997). The structure of urban landprices. Journal of Urban Economics, 41 (3), 321–336.Cook County, Office of the Assessor. (2005). Maps of Chicago properties.Retrieved October 15, 2005, from http://www.cookcountyassessor.com/filings/gis.aspDietrick, S., & Ellis, C. (2004). New urbanism in the inner city: A casestudy of Pittsburgh. Journal of the American Planning Association, 70 (4),426–443.Duany, A., Plater-Zyberk, E., & Speck, J. (2000). Suburban nation: Therise of sprawl and the decline of the American dream. New York: NorthPoint Press.Ellen, I. G., Schill, M., Susin, S., & Schwartz, A. E. (2001). Buildinghomes, reviving neighborhoods: Spillovers from subsidized constructionof owner-occupied housing in New York City. Journal of HousingResearch, 12 (2), 185–216.Eppli, M. J., & Tu, C. C. (1999). Valuing the new urbanism: Theimpact of the new urbanism on the prices of single-family homes. Washing-ton, DC: Urban Land Institute.Galster, G. (2002). An economic efficiency analysis of deconcentratingpoverty populations. Journal of Housing Economics, 11 (4), 303–329.Gatzlaff, D., & Haurin, D. (1997). Sample selection bias and repeat-sales index estimates. Journal of Real Estate Finance and Economics,14 (1/2), 33–50.Google. (2005). Map of Chicago. Retrieved October 15, 2005, fromhttp://www.maps.google.com

Guttery, R. S. (2002). The effects of subdivision design on housingvalues. Journal of Planning Literature, 17 (2), 262–331.Gyourko, J., & Rybczynski, W. (2001). Financing new urbanismprojects: Obstacles and solutions. Housing Policy Debate, 11 (3), 733–750.Hughes, W., Jr., & Turnbull, G. (1996). Uncertain neighborhood effectsand restrictive covenants. Journal of Urban Economics, 39 (2), 160–173.Jacobs, J. (1961). The death and life of great American cities. New York:Random House.Jargowsky, P. A. (1997). Poverty and place: Ghettos, barrios, and theAmerican city. New York: Russell Sage Foundation.Jargowsky, P. A. (2003). Stunning progress, hidden problems: Thedramatic decline of concentrated poverty in the 1990s (Working Paper,Brookings Institute Living City Census Series). Washington, DC:Brookings Institute.Leccese, M., & McCormick, K. (Eds.). (2000). Charter of the newurbanism. New York: McGraw-Hill.Morrow-Jones, H., Irwin, E., & Roe, B. (2004). Consumer preferencefor neotraditional neighborhood characteristics. Housing Policy Debate,15 (1), 171–202.Newman, O. (1972). Defensible space: Crime prevention through urbandesign. New York: Macmillan.Peiser, R. (1984). Does it pay to plan suburban growth? Journal of theAmerican Planning Association, 50 (3), 419–433.Ryan, B. (2002). The Suburbanization of the inner city: Urban housingand the pastoral ideal. Unpublished doctoral dissertation, MassachusettsInstitute of Technology, Cambridge.Ryan, B. (2006a). Privately-financed housing in distressed urban neighbor-hoods: Lessons from Detroit. Unpublished research report, Fannie MaeFoundation.Ryan, B. (2006b). Morphological change through residential redevelop-ment: Detroit, 1950–2000. Urban Morphology, 10 (1), 5–22.Sirmans, G. S., Macpherson, D., & Zietz, E. (2005). The compositionof hedonic pricing models. Journal of Real Estate Literature, 13 (1), 3–43.Sirmans, C. F., Turnbull, G., & Dombrow, J. (1997). Residential devel-opment, risk, and land prices. Journal of Regional Science, 37 (4), 613–628.Song, Y., & Knaap, G. (2003). New urbanism and housing values: Adisaggregate assessment. Journal of Urban Economics, 54 (2), 218–239.Speyrer, J. (1989). The effect of land-use restrictions on market valuesof single-family homes in Houston. Journal of Real Estate Finance andEconomics, 2 (2), 117–130.Steuteville, R. (1999). True urbanists: Research demonstrates that newurbanists are focusing substantial attention on infill planning anddevelopment. New Urban News, 4 (1), 6–12.Talen, E. (2001). Traditional urbanism meets residential affluence: Ananalysis of the variability of suburban preference. Journal of the AmericanPlanning Association, 67 (2), 199–216.Thorsnes, P. (2000). Internalizing neighborhood externalities: Theeffect of subdivision size and zoning on residential lot prices. Journal ofUrban Economics, 48 (3), 397–418.U.S. Department of Housing and Urban Development. (2000). HopeVI: Community building makes a difference. Washington, DC: Author.U.S. Department of Housing and Urban Development, & Congress forthe New Urbanism. (2000). Principles for inner city neighborhood design.Washington, DC: U.S. Department of Housing and Urban Development.Whyte, W. H. (1988). City: Rediscovering the center. New York: Doubleday.Wilson, W. J. (1996). When work disappears: The world of the new urbanpoor. New York: Knopf.Wyly, E. K., & Hammel, D. J. (1999). Islands of decay in seas ofrenewal: Housing policy and the resurgence of gentrification. HousingPolicy Debate, 10 (4), 711–771.

Ryan and Weber: Valuing New Development in Distressed Urban Neighborhoods 111

JAPA 73-1 07 Ryan fin 1/15/07 6:07 PM Page 111