breaking the chain: understanding the association between foreclosure intensity and neighborhood...

11
Breaking the chain: Understanding the association between foreclosure intensity and neighborhood satisfaction Misun Hur a, * , Yanmei Li b , Kathryn Terzano c a Department of Geography, Planning, and Environment, East Carolina University, NC, USA b School of Urban and Regional Planning, Florida Atlantic University, 777 Glades Road, SO 284, Boca Raton, FL 33431, USA c Wilson 203c, Department of Geography and Regional Planning, Westeld State University, 577 Western Avenue, Westeld, MA 01086, USA article info Article history: Available online Keywords: Neighborhood satisfaction Foreclosure intensity Chain effect Structural equation model abstract Extensive research has been conducted on the factors that cause foreclosure and the resulting neigh- borhood contagion effects. When residents see concentrated foreclosed homes in their neighborhood, their perceptions of their neighborhood are often negatively affected. Despite the potential psychological effects of foreclosures on residents, no research has looked at foreclosure intensity and impacts on other residents' satisfaction in the neighborhood. We used the 2004 Homeowner Satisfaction Survey and foreclosed property data from 2001 to 2004 in Franklin County, Ohio to explore the effect of foreclosure intensity on resident satisfaction. We used a structural equation model to capture the chain reaction of concentrated neighborhood disadvantage on foreclosure intensity and residents' neighborhood satisfaction. We found that the impacts of foreclosures are not limited to lenders and owners of mortgaged homes but extended to other residents living in the neighborhood. In this regard, we recommend housing recession recovery efforts must incorporate programs to alleviate neighbors' psychological distress. Published by Elsevier Ltd. Introduction The National Bureau of Economic Research declared a national recession from December 2007 to June 2009 (National Bureau of Economic Research, 2014). While the recession is technically over, the after-effects are still visible on various sectors of the economy and especially the mortgage nance markets. Statistics show that one in every 1170 homes (outof a total of 1,187,654 properties) in the U.S. received a foreclosure ling in February 2014 (RealtyTrac, 2014). Although foreclosure problems have become dominant since the subprime housing crisis in the late 2000s, the problem existed long before then. In early 2000, some Midwest states, such as Indiana and Ohio, experienced signicant increases in fore- closure lings, and Ohio ranked as one of the top states in fore- closures. Potential reasons that these states had high foreclosure rates include the prevalence of subprime lending and the loss of manufacturing jobs in the rust belt (Webb, 2009). The National Delinquency Survey of the Mortgage Bankers Association (MBA) 2004 dataset shows that the foreclosure rate in Ohio had been above the U.S. average since 1999. The Ohio foreclosure rate became the highest in all states in 2003 (with a rate of 2.9 percent compared to 1.2 percent nationally) and again in 2006 (with a rate of 3.38 percent compared to 1.19 percent nationally). Foreclosures have profound impacts on every aspect of our society, and the effects of foreclosures have reached far beyond lenders and borrowers. Factors causing foreclosures have been well researched, but studies about the neighborhood impacts of foreclosure, while diverse, have been limited to racial/ethnic transition (Baxter & Lauria, 2000), actual and potential criminal activities (Arnio, Baumer, & Wolff, 2012; Ellen, Lacoe, & Sharygin, 2013; Immergluck & Smith, 2006a; Zhang & McCord, 2014), default risks for surrounding bor- rowers in a neighborhood (Agarwal, Ambrose, Chomsisengphet, & Sanders, 2012; Chan, Gedal, Been, & Haughwout, 2013), and, most commonly, the spatial dependence of housing prices (Baumer, Arnio, & Wolff, 2013; Han, 2014; Immergluck & Smith, 2006b; Leonard & Murdoch, 2009; Lin, Rosenblatt, & Yao, 2009). The foreclosure effects have been especially prominent in economically vulnerable regions with concentrated disadvantages, such as high poverty rates, which often relate to the concentration of subprime lending in these areas and neighborhoods (Bocian, * Corresponding author. A-212 Brewster Building, Department of Geography, Planning, and Environment, East Carolina University, Greenville, NC 27858-4353, USA. Tel.: þ1 252 328 1270. E-mail addresses: [email protected] (M. Hur), [email protected] (Y. Li), kterzano@ westeld.ma.edu (K. Terzano). Contents lists available at ScienceDirect Applied Geography journal homepage: www.elsevier.com/locate/apgeog http://dx.doi.org/10.1016/j.apgeog.2015.01.004 0143-6228/Published by Elsevier Ltd. Applied Geography 58 (2015) 7e17

Upload: asu

Post on 04-Dec-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

lable at ScienceDirect

Applied Geography 58 (2015) 7e17

Contents lists avai

Applied Geography

journal homepage: www.elsevier .com/locate/apgeog

Breaking the chain: Understanding the association betweenforeclosure intensity and neighborhood satisfaction

Misun Hur a, *, Yanmei Li b, Kathryn Terzano c

a Department of Geography, Planning, and Environment, East Carolina University, NC, USAb School of Urban and Regional Planning, Florida Atlantic University, 777 Glades Road, SO 284, Boca Raton, FL 33431, USAc Wilson 203c, Department of Geography and Regional Planning, Westfield State University, 577 Western Avenue, Westfield, MA 01086, USA

a r t i c l e i n f o

Article history:Available online

Keywords:Neighborhood satisfactionForeclosure intensityChain effectStructural equation model

* Corresponding author. A-212 Brewster BuildingPlanning, and Environment, East Carolina UniversityUSA. Tel.: þ1 252 328 1270.

E-mail addresses: [email protected] (M. Hur), yli2westfield.ma.edu (K. Terzano).

http://dx.doi.org/10.1016/j.apgeog.2015.01.0040143-6228/Published by Elsevier Ltd.

a b s t r a c t

Extensive research has been conducted on the factors that cause foreclosure and the resulting neigh-borhood contagion effects. When residents see concentrated foreclosed homes in their neighborhood,their perceptions of their neighborhood are often negatively affected. Despite the potential psychologicaleffects of foreclosures on residents, no research has looked at foreclosure intensity and impacts on otherresidents' satisfaction in the neighborhood.

We used the 2004 Homeowner Satisfaction Survey and foreclosed property data from 2001 to 2004 inFranklin County, Ohio to explore the effect of foreclosure intensity on resident satisfaction. We used astructural equation model to capture the chain reaction of concentrated neighborhood disadvantage onforeclosure intensity and residents' neighborhood satisfaction. We found that the impacts of foreclosuresare not limited to lenders and owners of mortgaged homes but extended to other residents living in theneighborhood. In this regard, we recommend housing recession recovery efforts must incorporateprograms to alleviate neighbors' psychological distress.

Published by Elsevier Ltd.

Introduction

The National Bureau of Economic Research declared a nationalrecession from December 2007 to June 2009 (National Bureau ofEconomic Research, 2014). While the recession is technically over,the after-effects are still visible on various sectors of the economyand especially the mortgage finance markets. Statistics show thatone in every 1170 homes (out of a total of 1,187,654 properties) inthe U.S. received a foreclosure filing in February 2014 (RealtyTrac,2014). Although foreclosure problems have become dominantsince the subprime housing crisis in the late 2000s, the problemexisted long before then. In early 2000, some Midwest states, suchas Indiana and Ohio, experienced significant increases in fore-closure filings, and Ohio ranked as one of the top states in fore-closures. Potential reasons that these states had high foreclosurerates include the prevalence of subprime lending and the loss ofmanufacturing jobs in the rust belt (Webb, 2009). The National

, Department of Geography,, Greenville, NC 27858-4353,

[email protected] (Y. Li), kterzano@

Delinquency Survey of the Mortgage Bankers Association (MBA)2004 dataset shows that the foreclosure rate in Ohio had beenabove the U.S. average since 1999. The Ohio foreclosure rate becamethe highest in all states in 2003 (with a rate of 2.9 percentcompared to 1.2 percent nationally) and again in 2006 (with a rateof 3.38 percent compared to 1.19 percent nationally).

Foreclosureshaveprofound impacts oneveryaspectof our society,and the effects of foreclosures have reached far beyond lenders andborrowers. Factors causing foreclosures have been well researched,but studies about the neighborhood impacts of foreclosure, whilediverse, have been limited to racial/ethnic transition (Baxter& Lauria,2000), actual and potential criminal activities (Arnio, Baumer, &Wolff, 2012; Ellen, Lacoe, & Sharygin, 2013; Immergluck & Smith,2006a; Zhang & McCord, 2014), default risks for surrounding bor-rowers in a neighborhood (Agarwal, Ambrose, Chomsisengphet, &Sanders, 2012; Chan, Gedal, Been, & Haughwout, 2013), and, mostcommonly, the spatial dependence of housing prices (Baumer, Arnio,& Wolff, 2013; Han, 2014; Immergluck & Smith, 2006b; Leonard &Murdoch, 2009; Lin, Rosenblatt, & Yao, 2009).

The foreclosure effects have been especially prominent ineconomically vulnerable regions with concentrated disadvantages,such as high poverty rates, which often relate to the concentrationof subprime lending in these areas and neighborhoods (Bocian,

M. Hur et al. / Applied Geography 58 (2015) 7e178

Ernst, & Li, 2008; Immergluck & Smith, 2005), and a lack ofattractive amenities for reinvestment (Johnson, Turcotte, &Sullivan, 2010). Sampson, Raudenbush, and Earls (1997) showedthat concentrated disadvantages are negatively related to neigh-borhood collective efficacy and residential stability. Concentrateddisadvantages thus indirectly (or directly to certain extent) relate tothe concentration of foreclosures in certain neighborhoods.When aforeclosed house becomes vacant and boarded up, it may trigger acontagion effect, causing discontent and lack of motivation tomaintain the properties by other residents in the same neighbor-hood. The physical disorder associated with foreclosed homes alsodeters the interests fromhomebuyers or real estate investors to buyhomes in the neighborhood, thus further degrading the physicalquality of the neighborhood. This in turn influences residents'negative perceptions of their neighborhooddincreasing their fearof crime and lowering their neighborhood satisfaction and sense ofcommunity (Hur & Nasar, 2014). Therefore, according to the socialdisorganization theory, foreclosure-related residential turnovercreates neighborhood instability, and thus causing neighborhooddecline and decay (Schildkraut & Mustaine, 2013). In addition toresidential instability, the presence of concentrated disadvantagesis one of the major factors explaining social disorder and disorga-nization associated with foreclosures. Foreclosures create all thesefactors contributing to social disorganization, which results in thedestruction of neighborhood social networks and then residentialdissatisfaction and psychological distress.

It is imperative to understand the chain reaction of foreclosureson neighborhood decline in order to design effective public policiesthat address the issues related to foreclosure and neighborhoodstabilization. Our interest in the research is to focus on foreclosureand its effects on other residents' (not mortgage borrowers') per-ceptions in the neighborhood. Two our hypotheses are: 1) an in-crease of foreclosed homes in a neighborhood is negativelyassociated with residents' neighborhood satisfaction, and 2) resi-dents become more dissatisfied as the intensity of foreclosure in-creases through closer distance. This research contributes to thecurrent literature on foreclosure and neighborhoods from the angleof the cognitive perception of homeowners living in proximity toforeclosed properties as well as on the methodology utilizing astructural equation model with combined subjective survey dataand objective geographic information data. Traditional neighbor-hood satisfaction theory indicates that real and perceived neigh-borhood characteristics and household characteristics determineneighborhood satisfaction. Our research adds foreclosures as one ofthese real neighborhood characteristics determining neighborhoodsatisfaction, in response to the increasing foreclosure trend in theU.S. in the 2000s. Since the aggregated neighborhood disadvan-tages will be used in the model, individual household characteris-tics are treated as omitted variables to avoid the reflectionproblems in neighborhood effect research. Using a path model thisresearch will more accurately reveal the interrelationship amongconcentrated disadvantages, building density, neighborhood in-teractions, foreclosure intensity, and satisfaction with neighbor-hoods. Our research also uses a multiple array of itemizedneighborhood satisfaction variables, which are proven to be relatedto foreclosure intensity of a neighborhood, to measure neighbor-hood satisfaction. This will contribute greatly to the literaturemeasuring neighborhood satisfaction on specific characteristics e

real or perceived e of a neighborhood.

Mortgage foreclosure and neighborhood satisfaction

There has been rich research on neighborhood satisfaction.Theories of neighborhood satisfaction are diverse due to its com-plex and multidimensional nature (Amerigo & Aragones, 1997;

Marans & Rodgers, 1975; Weidemann & Anderson, 1985). Howev-er, much of the research has focused on how neighborhood char-acteristics or individual attributes contribute to residents'neighborhood satisfaction. Empirical studies have tried to ascertainfactors and their associations with neighborhood satisfaction. It iswidely believed that aggregated neighborhood attributes predict aresident's overall neighborhood satisfaction (Marans & Rodgers,1975). Research has suggested affective factors on neighborhoodsatisfaction that broadly include aesthetics (Hur & Morrow-Jones,2008), physical characteristics (Kim & Kaplan, 2004; Patterson &Chapman, 2004), socioeconomic characteristics (Bruin & Cook,1997; Miller, Tsemberis, Malia, & Grega, 1980), and sociodemo-graphic characteristics (Alvi, Schwartz, DeKeseredy, & Maume,2001; Bruin & Cook, 1997; Westaway, 2007). More importantly,homeownership is believed to be a prominent factor contributingto neighborhood satisfaction (Galster, 1987). Homeownership mo-tivates residents to maintain their residences and encourages themto participate in neighborhood organizations, events, and activities,thereby creating a sense of community and neighborhood stability(Rohe & Stewart, 1996).

Despite numerous attempts to understand the characteristicsand underlyingmechanisms of neighborhood satisfaction, only littleresearch in the literature has examined the effects of mortgageforeclosures and residents' neighborhood satisfaction. Notably,Batson andMonnat (2014) found an association between foreclosurerate and quality of life and neighborhood satisfaction. But theycalculated foreclosure rate based on the census tract, which does notreflect neighborhood characteristics. Research by Grinstein-Weisset al. (2011) and Carswell, James, and Mimura (2009) addressedconnections between foreclosures and residential satisfaction andthe role of counselors in avoiding foreclosure, and their relationshipsto neighborhood satisfaction. These studies were, however, focusedmore on identifying future research areas and mechanisms on howto reduce foreclosure. What seems to be glaringly missing in theliterature is an empirical analysis that examines the relationshipbetween foreclosure intensity and residential satisfaction in theneighborhood. In this paper, we operationalize foreclosure intensityby identifying the ratio of foreclosed homes in a given geographicarea among total parcels in the area.

Neighborhood satisfaction is operationalized through a multi-dimensional scale that includes a set of individual characteristicsand perspectives measured in a Likert scale. Selected neighborhoodsatisfaction variables are directly or indirectly related to the nega-tive consequences and impacts of foreclosures. These satisfactionvariables include satisfactionwith safety from crime, neighborhoodcleanliness, homeownership rate, general appearance, proximity toproblem areas, and housing density. Vacant, vandalized propertieswithout proper maintenance often increase the perception ofcrime, decrease cleanliness, and affect the general physicalappearance of the neighborhood. These effects are often accentu-ated if the foreclosure is highly concentrated in certain areas(problem areas). Higher housing density might intensify the con-centration of foreclosures due to the contagion effect of fore-closures (e.g. Immergluck & Smith, 2006a, 2006b, among others).

Numerous extant studies indicate that foreclosures are spatiallyconcentrated in lower income minority neighborhoods with ahigher percentage of poverty rate, unemployment rate and othersocio-demographic disadvantages. The concentration of thesenegative neighborhood demographic and economic characteristicsis often referred to as concentrated disadvantages. Neighborhoodswith concentrated disadvantages were often targeted by subprimelending (Bocian et al., 2008; Grover, Smith, & Todd, 2008). Sub-prime lending is proven to be one of the most significant factorscontributing to residential mortgage foreclosure due to its high-cost nature and its role to increase the housing cost burden of

1 Franklin County, Ohio, is composed of 25 smaller municipalities, includingcities, townships, and villages. The City of Columbus, located at the center (green inFig. 1, in the web version), covers the majority of the area. Most of the inner suburbswere established in early 1900 and are located inside the I-270 outer belt highway(yellow in Fig. 1, in the web version). The outer-ring suburbs are much more recent,and their development patterns go towards the northwest, the north, the northeast,and the west (purple in Fig. 1, in the web version).

M. Hur et al. / Applied Geography 58 (2015) 7e17 9

lower-income households (Gramlich, 2007; Immergluck & Smith,2005; Lucy, 2010). However, during the housing crisis, unemploy-ment associated with the recession might also become one of themost important determinants of foreclosures.

One potential outcome in a neighborhood with numerousforeclosures is the concentration of abandoned vacant properties,whichmay provoke actual or perceived crimes, thus contributing tophysical and social disorder in the neighborhood. When residentsin the neighborhood sense the disorder, they perceive the neigh-borhood to be unsafe (Wallace, Hedberg, & Katz, 2012). As the fearof crime increases residents tend to restrict their neighborhood lifeand interaction, thus breaking the cohesion of the social network inthe neighborhood. Therefore social disorder associated with fore-closures decreases neighborhood satisfaction, reducing the resi-dents' willingness to help their neighbors and the motivation tomaintain their properties. Increased social disorder (regardless if itis real or perceived) and decreased connection and interactionwithneighbors decrease neighborhood satisfaction of the residents(Batson & Monnat, 2014; Drassopoulos, Batson, Futrell, & Brents,2012; Wallace et al., 2012). Decreased neighborhood satisfactionincreases the tendency to move and therefore exacerbates resi-dential instability resulting from foreclosures. Wallace et al. (2012)found that prior to the housing crisis, foreclosures were related toshort-term social disorder of a neighborhood, yet the finding didnot hold during the recession. This suggests the methodology ofusing data prior to the housing crisis to explore the relationshipbetween foreclosures and neighborhood satisfaction. Although thestudies about the impact of foreclosures on social disorder anddisorganization yielded mixed results, research has repeatedlyconfirmed that perceived upkeep and safety from crime positivelyimpact neighborhood satisfaction (Brown, Perkins, & Brown, 2004;Hur & Nasar, 2014; O'Brien & Wilson, 2011).

There has not been any direct and systematic research on theimpact of concentrated foreclosures on public health. However, theconnection between foreclosure intensity and psychologicaldistress is embedded within the literature on disadvantagedneighborhoods. Previous research on disadvantaged neighbor-hoods, which are characterized by ongoing issues with socialproblems (e.g., unemployment, teen pregnancy, crime, and druguse) as well as physical problems (notably, abandoned houses), hasfound that these types of neighborhoods can be psychologicallydistressing (Hill & Angel, 2005). Residents of these neighborhoodsexhibit much higher levels of psychological distress, includingdepression, than residents of more advantaged neighborhoods(Kim, 2010; Latkin & Curry, 2003). Another study found a link be-tween higher levels of stress and higher levels of neighborhoodinstability, as defined by rates of homeownership and length ofresidency in the neighborhood (Boardman, 2004).

Additionally, social disorder, mistrust, fear of crime and theconsequent dissatisfactionwith the neighborhood will unavoidablydecrease resident social capital and contribute to deterioratedpsychological and mental health of the residents (Bennett,Scharoun-Lee, & Tucker-Seeley, 2009). Foreclosed homes oftenbring negative attention to a neighborhood, including auction ad-vertisements, for-sale signs, REO (Real-Estate Owned) notices,dilapidated houses and lots, boarded-up houses, and potentialvandalism associated with such properties. Research indicates thathousing conditions, regardless of tenure, influence households'self-esteem and life satisfaction among low-income people; whilehomeownership alone is only related to increased life satisfaction(Rohe & Stegman, 1994). Therefore, the deteriorated housing con-ditions will decrease self-esteem and life satisfaction of households.The contagion effect of foreclosures on adjacent properties mightalso increase the anxiety of surrounding households for the fear ofincreased foreclosure risks. All of these physical and perceptional

outcomes associated with concentrated foreclosures may nega-tively influence the neighborhood satisfaction of other residents,thus increasing psychological stress of all the residents in theneighborhood.

Materials and methods

Research procedure

In this study, we used the 2004 Homeowner Satisfaction Surveyconducted by the Ohio State University Survey Research Center inFranklin County, the central county of Ohio, which includes theColumbus metropolitan area. Franklin County has a diverse anddynamic housingmarket, yet the extent of foreclosures is much lessprominent compared to Cuyahoga County, where the City ofCleveland is located. Based on the metropolitan delinquency andforeclosure data compiled by Foreclosure-Response.org, over theyears the Columbus MSA consistently showed average delinquencyand foreclosure rates. Therefore Franklin County is a good repre-sentation of an average U.S. metropolitan area affected greatly byforeclosures, yet not significant enough to rank among the topforeclosure markets. The survey data were merged with the U.S.Census 2000 demographic data, the HomeMortgage Disclosure Act(HMDA) data (2004e2006), and foreclosure data (2001e2004)obtained from the Franklin County Sheriff's Department. Thereforethe data used in this research is before the housing foreclosurecrisis in late 2000s. The foreclosure intensity was much weakerthan in the Great Recession. However, Wallace et al. (2012) indicatethat foreclosures contributed to social disorders prior to thehousing crisis, not during the crisis. Our research will provideanother empirical study about whether and how foreclosures relateto neighborhood satisfaction prior to the recession. This might helpus to compare the same type of relationship during or after therecession.

The 2004 Homeowner Satisfaction Survey included questionson homeowners' degree of satisfaction with a range of neighbor-hood characteristics, as well as preferred conditions for resident'sfuture housing such as, housing type, neighborhood layout, density,land use, school quality, neighborhood safety, housing price, andyear built. In the spring of 2004, the survey team randomly selecteda sample (2600 properties) from a list of homeowners included inthe Franklin County Tax Auditor's database. The survey team thenidentified the name and address of each property owner andmailed out the Homeowner Satisfaction Survey. A pair of tickets toan Ohio State University home football game was raffled off to therespondents who returned their surveys by a certain date. Theresearch team asked an adult who presently resides in the house-hold to answer to the survey. About 32.2 percent of the originalsample responded (837 respondents). Fig. 1 shows the respondentdistributions in Franklin County.1

For data on home foreclosure, we used the sheriff's foreclosuresales data to calculate foreclosure intensity. The State of Ohio fol-lows judicial foreclosure procedures. When a mortgage is in defaultand the borrower fails to make the back payment(s) or requests ashort-sale, mortgage lenders may choose to file a foreclosure withthe court. If the case cannot be resolved and the borrower continuesto be in default, the court can order a public auction of the property.

Fig. 1. Respondent locations of Homeowner Satisfaction Survey, Franklin County, Ohio.

M. Hur et al. / Applied Geography 58 (2015) 7e1710

Properties sold at auction or returned to the lender (if the highestbid is not accepted) are recorded as sheriff's deeds in the deedtransfer records. Since the judicial foreclosure process is longerthan the non-judicial process, taking several months to a few years,around 30 percent of foreclosed properties complete the fore-closure process and are recorded as sheriff's deeds in FranklinCounty each year. The rest of the foreclosure filings either roll overto the next year for further processing or are resolved before goingto public auction. We used foreclosed homes recorded as sheriff'sdeeds for this study. These recorded foreclosed homes are oftensold at a deep discount at the auction or as REO (Real Estate Owned)

Fig. 2. Foreclosed homes in Franklin County, Ohi

properties. We believed that these recorded foreclosed homesserve as an appropriate proxy for foreclosed homes and areappropriate for our use.

We retrieved the sheriff's deed transfer data recorded in theCounty Recorder's office for the period 2001e2004. A total of 5460foreclosed homeswere included in the analysis. Fig. 2 illustrates thetotal number of foreclosed homes in the census tract in quartiles.Using GIS, we mapped the respondents of the Homeowner Satis-faction Survey and the locations of foreclosed homes. We thencalculated the foreclosure ratio within radial distances from surveyrespondents' locations.

o (Sheriff's Deed transfer data, 2001e2004).

Fig. 3. The relationship between respondent location and foreclosure intensity (concept map).

2 Long and Perkins (2003) suggested, “When theoretical and empirical evidenceexists for a multidimensional construct, confirmatory factor analysis is the appro-priate procedure for confirming the fit of the theoretical structure” (p. 280).Confirmatory factor analysis (CFA, theory driven) differs from explanatory factoranalysis (EFA, data driven) (Anderson & Gerbing, 1988). EFA has been used toexplore the possible underlying factor structure of a set of manifest variableswithout imposing a preconceived structure on the outcome. In contrast, CFA is usedto verify the factor structure of a set of manifest variables. CFA can test hypothesesabout a particular factor structure and allows researchers to assess the reliabilityand validity of constructs and indicators (Hatcher, 1994).

M. Hur et al. / Applied Geography 58 (2015) 7e17 11

The literature suggests that property values are reducedbasedonthe number of foreclosures locatedwithin a given distance. Leonardand Murdoch's (2009) observations on 2006 home sales in DallasCounty, Texas, found that a selling price discount for a home withinapproximately 1/8-mile of a foreclosure was five times greater thana selling price discount for a homewithin approximately 1/4-mile,1/2-mile, or 1-mile of a foreclosure. Immergluck and Smith (2006b)also supported the 1/8-mile distance to measure the effects offoreclosures on property value change. They found that a foreclosedpropertywithin the 1/8-mile distance is associatedwith a .9 percentdecrease in property values in the radius.

This study borrows the established spatial measurements/dis-tance of the impact of foreclosures from other home value impactliterature. We applied four radial distancesd1/8-mile, 1/4-mile, 1/2-mile, and 1-miledto capture the foreclosure ratio within eachradial distance in relationship to the respondent's home. Fig. 3 il-lustrates our concept to identify the data used in this study.

Assuming that the concentration of neighborhooddisadvantagesmay be associated with higher foreclosure rates in a neighborhood,we added three neighborhood disadvantage concentration featuresto the research using 2000 Census and the Home MortgageDisclosure Act (HMDA) data (2004e2006): percent of populationbelow the poverty line, percent of unemployed population, andpercent of high-cost homepurchase loans (subprimemortgages) for1e4 unit housing among all mortgages outstanding, all at the samespatial scale, the census tract. HMDA defines high-cost loans asthose with annual percentage rates (APR) exceeding the yield oncomparable Treasury securities by three or more percentage pointsfor first-lien loans and about five percentage points above theTreasury rate for second-lien loans (The Federal Reserve, 2006). Alldata were joined in ArcGIS and then imported into SAS for analysis.

Residents' neighborhood satisfaction depends on many closelyrelated factors, which work not only as predictor variables but alsoas mediating or moderating variables. Much neighborhood satis-faction research has used regression analysis. However, regressionanalysis cannot capture this kind of complexity where there are

moderating effects. To understand the complex structure ofneighborhood satisfaction, we used structural equation modeling(SEM). SEM is a statistical method that makes apparent the com-plex relationships between multiple independent and multipledependent variables (Hur & Nasar, 2014). In contrast to regressionanalysis, which only shows the “direct” impact between indepen-dent variables and the dependent variable, the SEM model showsthe indirect effects among structural variables.

A SEM model consists of two components: the measurementportion (referred to as the measurement model or the confirmatoryfactor analysis, CFA2), which describes the relationships betweenthe latent factors and their indicator variables, and the structuralportion of the model (referred to as the structural model), whichdescribes the predicted relationships between the latent factors(Hatcher, 1994). To test both the measurement and structural por-tions of the model, we followed Anderson and Gerbing's (1988)two-step modeling process, testing the measurement model firstand then the structural model.

The analysis

Descriptive characteristics of dataTable 1 compares the demographic characteristics of survey

respondents to residents of Franklin County. The demographiccharacteristics showed that our sample tended to be older, more

Table 1Demographic characteristics of survey respondents to Franklin County residents.

Homeownersurvey respondents(n ¼ 837)

Franklin Countypopulation in2000 Census

Residency:City of Columbus

51.9%Inner-ring suburbs

12.3%Outer-ring suburbs& exurbs 35.8%

Gender:Male

53.6%48.6%

Female44.8%

51.4%

Missing1.6%

Marital status:Married

70.0%47.9%

Single12.2%

33.1%

Other17.0%

19.0%

Missing.8%

Age:Median age 49 years old 32.5 years old

*Age ranges used inthe survey and thecensus data differ.

<20 years old .2% <20 years old 28.2%20e29 years old 4.2% 20e24 years old 8.6%30e39 years old 19.5% 25e34 years old 17.1%40e49 years old 25.4% 35e44 years old 16.2%50e59 years old 26.4% 45e54 years old 12.8%60e69 years old 10.0% 55e59 years old 4.1%70e79 years old 5.4% 60e64 years old 3.2%>¼80 years old 2.0% 65e74 years old 5.3%Missing 6.8% >¼75 years old 4.5%

Race:Non-Hispanic White 85.7% 75.5%African American 6.3% 17.9%Asian .6% 3.1%Latino .6% 2.3%Other .8% 1.7%Missing 6.0%

Education:Less than high school 1.1% 14.3%High school graduates 20.5% 27.1%Some college 5.3% 21.3%Associate degree 13.0% 5.5%Bachelor's degree 33.7% 21.2%Master's orprofessional degree

23.3% 10.7%

Missing 3.1%

M. Hur et al. / Applied Geography 58 (2015) 7e1712

settled, and more educated, with a higher proportion of non-Hispanic Whites than Franklin County residents. Although this isconsistent with previous studies indicating that minorities, andparticularly African Americans, are less likely to participate insurveys (Johnson, O'Rourke, Burris, & Owens, 2002), generaliza-tions from the research should reflect that fact.

SEM analysisWe included a total of 17 manifest variablesdnine variables

from the survey and eight variables from secondary data and GIScalculations. Table 2 presents a lists of variables utilized in thisresearch and their descriptive statistics. All variables except Build-ing density were grouped into four latent factors: Concentrateddisadvantages, Community interactions, Foreclosure intensity, andNeighborhood satisfaction (see Table 2 for variable groupings offactors). Therewere 674 sets of completed data, leaving 163missingdata. Deleting responses with missing values results in lower sta-tistical power (with lower n) and also may cause biases by ignoringinformation (for other variables). Instead, we used the simpleimputation method to replace missing values with mean values.

The SAS System's CALIS (Covariance Analysis and Linear Struc-tural Equations) procedure using the maximum likelihood methodof parameter estimation was applied to analyze the data. We usedthe covariance matrix for analysis with the data size of 837.3

Results

The measurement model (confirmatory factor analysis, CFA)

The measurement model tested the fit of the theoretical modelwith all 17 variables included and then suggested applying modi-fications to achieve an acceptable model fit. Hatcher (1994) rec-ommends identifying parameters to drop rather than to add newones because as it is generally safer when modifying a model. Wedropped three suggested variablesdforeclosure ratio within 1/2-mile radius, foreclosure ratio within 1-mile radius and high-costmortgage. The model fit indexes suggested that our measurementmodel fell in the reasonable error of approximation with the RootMean Square Error of Approximation (RMSEA ¼ .05), the Compar-ative Fit Index (CFI ¼ .97), the Bentler Bonett Index or Normed FitIndex (NFI ¼ .96), the Non-normed Fit Index (NNFI ¼ .96), and theGoodness of Fit Index (GFI ¼ .96).4

All manifest variables showed significantly high factor loadingsto their latent factorsdgreater than .50 (Table 3).5 The Foreclosureintensity factor was well measured by both Foreclosure ratio within1/8-mile radius and Foreclosure ratio within 1/4-mile with thestrength of association of .88 and .99, respectively. This suggeststhat residents are more concerned with foreclosed homes closer totheir residence. It also seems to suggest that the 1/2 mile-radius isthe threshold distance beyond which homeowners' concern aboutforeclosed home seems less critical. The Neighborhood satisfaction

Income:*Income ranges used

in the survey and thecensus data differ.

Below $20K 3.3% Less than $10K 8.9%$20Ke40K 13.4% $10Ke14.999K 5.6%$40e60K 18.5% $15Ke24.999K 12.3%$60Ke80K 17.7% $25Ke34.999K 13.4%$80Ke100K 14.1% $35Ke49.999K 16.9%$100Ke120K 10.6% $50Ke74.999K 20.9%$120Ke140K 5.9% $75Ke99.999K 10.5%$140Ke160K 3.8% $100Ke149.999K 7.6%$160Ke180K 2.5% $150Ke199.999K 2.0%$180Ke200K 1.4% $200K or more 2.0%Over $200K 3.7%Missing 5.0%

Live with children 19.0% 33.5%

3 If the input step includes both a correlation matrix and standard deviations, theSAS system (CALIS procedure with COVARIANCE option) will use these to create acovariance matrix. Without this option, the analysis is performed on a correlationmatrix by default (Kline, 2005).

4 Schreiber, Stage, King, Nora, and Barlow (2006) suggest the RMSEA values of .01,.05, and .08 to indicate excellent, good, and mediocre fit, respectively. For CFI, NFI,NNFI, and GFI, a value below .90 is a poor fit, between .90 and .95 is consideredmarginal, and above .95 is good.

5 Hair, Black, Babin, and Anderson (2009) suggest “factor loadings in the range of±.30 to ±.40 are considered to meet the minimal level for interpretation of struc-ture; loadings ±.50 or greater are considered practically significant; and loadingsexceeding ±.70 are considered indicative of well-defined structure” (p. 117).Hatcher (1994) also recommends “the standard factor loadings .60 or higher to beconsidered as moderately large” (p. 295).

Table 2Variables, factors, brief descriptions and summary statistics.

Group Variable Description Mean SD Min. Max.

Concentrateddisadvantages

Below poverty line Percent of population below thepoverty line

6.07 8.11 0 65.04

Unemployed Percent of unemployed civilianswho were actively looking for jobs

3.18 3.26 0 31.68

High-cost mortgage Percent of high-cost mortgage 23.02 15.23 3.88 71.43

Communityinteractions

Talking Frequency of talking with neighbors(ordinal scale)

3.42 1.01 1 5

Do a favor Frequency of doing favors for theneighbors (pick up mail when theyare out of town and let them borrowsomething; ordinal scale)

2.78 .83 1 5

Neighborhoodactivity/event

Frequency of participation inneighborhood activity/event(ordinal scale)

1.61 .64 1 5

Foreclosureintensity

Foreclosure ratiowithin 1/8-mile radius

Ratio of foreclosed homes per totalhousing within 1/8-mile radius

.01 .02 .00 .15

Foreclosure ratiowithin 1/4-mile radius

Ratio of foreclosed homes per totalhousing within 1/4-mile radius

.01 .01 .00 .11

Foreclosure ratiowithin 1/2-mile radius

Ratio of foreclosed homes per totalhousing within 1/2-mile radius

.01 .01 .00 .07

Foreclosure ratiowithin 1-mile radius

Ratio of foreclosed homes per totalhousing within 1-mile radius

.01 .01 .00 .05

Perceivedneighborhoodsatisfaction

Safety from crime Level of satisfaction with safety fromcrime in the neighborhood (7-Likert scalewhere 1 ¼ very dissatisfied and 7¼ very satisfied)

5.14 1.57 1 7

Cleanliness Level of satisfaction with cleanliness(free from litter) in the neighborhood(7-Likert scale: same as above)

5.68 1.46 1 7

Ratio of homeowner/renter Level of satisfaction with the ratio ofhomeowner/renter in the neighborhood(7-Likert scale: same as above)

5.44 1.56 1 7

General appearance Level of satisfaction with general physicalappearance of the neighborhood(7-Likert scale: same as above)

5.63 1.38 1 7

Proximity to problem areas Level of satisfaction with proximity toproblem areas such as railroad tracks ora drug house (7-Likert scale: same as above)

5.06 1.86 1 7

Housing density Level of satisfaction with density ofhousing within the neighborhood(7-Likert scale: same as above)

5.13 1.40 1 7

Building density Ratio of building footprint per acreon census tract

.11 .05 .01 .29

M. Hur et al. / Applied Geography 58 (2015) 7e17 13

factor is best measured by the Satisfaction with general appearance(factor loading ¼ .87) and Satisfaction with cleanness (factor loading¼ .86); the Concentrated disadvantages factor is best measured bythe Below poverty line (factor loading ¼ .90); and the Communityinteractions factor is best measured by the Do a favor and Talking(factor loading for both variables ¼ .75).

Table 3 also shows the validity and reliability assessment resultsof the measurement model. All validity and reliability indexessupport that our measurement model successfully captured thehypothetical constructs with significant components.6

6 The fact that all t tests are significant in our model shows that all indicators areeffectively measuring the same construct (Anderson & Gerbing, 1988). Theconvergent validity is assessed by the t tests for the factor loadings. The mea-surement model also shows that most factors and variables have higher values onreliability indexes. Although the indicator reliabilities of several Neighborhoodsatisfaction variables are low, the composite reliability of the Neighborhood satis-faction factor is high. Hatcher (1994) suggests .50 as the acceptable level of theindicator reliability and .60 or .70 as being minimally acceptable level of reliabilityfor composite reliability. The variance-extracted estimates also showed estimateslarger than or close to .5, which shows that variance due to measurement error issmaller than the variance captured by the factors (Hatcher, 1994).

The SEM model

Fig. 4 describes our nonstandard SEM analysis results graphi-cally. We illustrate the measurement component with thin linesand the structural component with bolded lines. The associationsbetween latent factors are displayed on single-headed straight ar-rows with standardized path coefficients, indicating the strength ofassociations. The covariance among exogenous variables is dis-played on a double-headed curve arrow with an estimate. R-squarevalues are located outside of each latent factor.

The structural equations of our model may be read off the pathdiagram:

y1i ¼ g10 þ g11x1i þ z1i

y2i ¼ g20 þ g21x1i þ g22x2i þ z2i

y3i ¼ g30 þ g31x1i þ g32x2i þ b31y1i þ b32y2i þ z3i

Table 3Properties of the model (validity and reliability).

Factor/indicator Standardized factor loading Convergent validity* (t) Reliability Variance-extracted estimate

Composite Indicator

Concentrated disadvantages .83 .70Below poverty line .90 51.74 .81Unemployed .77 39.55 .60

Community interactions .73 .48Talking .75 25.80 .57Do a favor .75 25.78 .56Neighborhood activity/event .57 18.77 .32

Foreclosure intensity .93 .88Foreclosure ratio in 1/8-mile radius .88 62.25 .77Foreclosure ratio in 1/4-mile radius .99 75.67 .98

Perceived neighborhood satisfaction .88 .56Safety from crime .74 42.20 .55Cleanliness .86 74.73 .74Ratio of homeowner/renter .68 32.86 .46General appearance .87 75.75 .75Proximity to problem areas .64 28.94 .41Housing density .66 31.08 .44

Note: * denotes that all t-values were significant at p < .001.

M. Hur et al. / Applied Geography 58 (2015) 7e1714

� Exogenous variables are represented by x's; endogenous variablesby y's; and disturbances by z's.

� g's are used for structural parameters relating an endogenous to anexogenous variable.

� b's are used for structural parameters relating one endogenousvariable to another.

Our SEM model reveals several direct associations betweenstructural variables. It shows a strong positive association betweenthe Concentrated disadvantage and Foreclosure intensity factors,which suggests that a neighborhood with higher concentrateddisadvantages (i.e., with a higher unemployment rate and a higherproportion of residents below the poverty line) is more likely tohave a higher foreclosed ratio.

Fig. 4. The SEM model of Concentrated disadvantages, Building density, Foreclosure intensity, Cthat the association is significant at p < .05; a single asterisk (*) denotes that the associatiop < .001.

The Neighborhood satisfaction had a strong negative associationwith the Concentrated disadvantages (�.39) and Foreclosure in-tensity (�.30) but, surprisingly, had weak associations with theCommunity interactions (.09) and Building density (�.09). The find-ings suggest that residents are dissatisfied when there is a higherdegree of concentration of neighborhood disadvantages and ahigher degree of foreclosed homes close to their homes.

The direction, magnitude, and relationship between factors aresubject to change when we take the indirect associations intoconsideration. When we combine both direct and indirect paths,we need to reread the findings (the total strength of associationchanges). Table 4 summarizes the relationships between factorswith direct, indirect, and total effects in our SEM. For example, theSEM suggests an indirect path between the Concentrated

ommunity interactions, and Neighborhood satisfaction. Note: double asterisks (**) denoten is significant at p < .01; and no asterisk denotes that the association is significant at

Table 4Direct, indirect, and total effects between structural variables (standard coefficientin Fig. 4).

From factor To factor Directeffect

Indirect effect Totaleffect

Concentrateddisadvantages

Foreclosureintensity

.62 .62

Building density Foreclosureintensity

�.13 �.13

Building density Communityinteractions

.10 .10

Concentrateddisadvantages

Neighborhoodsatisfaction

�.39 (.62 � �.30) �.58

Building density �.09 (.10 � .09)þ (�.13 � �.30)

�.04

Foreclosure intensity �.30 �.30Community interactions .09 .09

M. Hur et al. / Applied Geography 58 (2015) 7e17 15

disadvantages and Neighborhood satisfaction factors through theForeclosure intensity.When direct and indirect effects are combined,the total strength of association becomesmuch stronger (from�.39to �.58). The path model suggests that perhaps residents experi-ence a stronger dissatisfaction with the neighborhood not only bythe high foreclosure ratio (as one phenomenon) but also by itsdetrimental weighted social and economic impacts of having moreneighbors being unemployed and falling below the poverty line.

Conclusions

There have been massive efforts to address the housing fore-closure crisis since the national recession. The impacts of fore-closure have been a hot research topic to researchers as well. Thisstudy contributes a different perspective to the literature on theimpacts of foreclosure on neighborhoods by focusing on residentswhose homes were not subject to mortgage foreclosures. Wedemonstrated how the intensity of foreclosure in a neighborhoodrelates to neighborhood satisfaction.

The empirical findings support our hypothesesdwhen there is ahigh concentration of foreclosed homes in a neighborhood, resi-dents showed higher dissatisfaction with their neighborhood. Thisfinding also highlights that the victims of the housing foreclosurecrisis are not limited to those who are in delinquency but it alsoincludes those who live in neighborhoods with a concentration offoreclosed homes, most notably those living within a radius of lessthan 1/2 mile. Themodel depicts the chain reaction of concentratedneighborhood disadvantages; that is, when there are concentrateddisadvantages in the neighborhood, the foreclosure intensity in thearea is also high and thus residents are dissatisfied with theirneighborhood. The strength of the association was notably high.

Building density is an interesting variable in the model. It isassociated positively with the Community interactions, which ex-plains that people in high dense areas have more opportunities tointeract with neighborsdtalking, offering a favor, and participatingin neighborhood eventsdand eventually results in the positiveneighborhood satisfaction among residents. However, when thephysical characteristics are considered directly, Building density hasa weak negative association with Neighborhood satisfaction. TheBuilding density variable also has a negative association with Fore-closure intensity, indicating that the area with higher foreclosureintensity is not necessarily a dense urban neighborhooddeven lessdense suburbs experience higher foreclosure intensity.

The findings highlight one important aspect: the impacts ofconcentrated disadvantages and foreclosures are not limited tolenders and borrowers of mortgaged homes but it extends to otherresidents living in the neighborhood. A high ratio of foreclosedhomes in a close proximity to resident can negatively affect his/her

neighborhood satisfaction in terms of cleanliness, general appear-ance, safety from crime, ratio of homeowners/renters, proximity toproblem areas, and housing density. This is valuable information forplanners and policymakers who are planning for the housingrecession recovery. Such efforts should incorporate programs toalleviate neighbors' psychological distress. In recent years, banksandmunicipalities have chosen to bulldoze some foreclosed housesto avoid further dilapidation of the physical appearance of prop-ertiesdlot and buildingdand to deter squatters and criminals (CBSNews 60 Minutes, 2011; Gandel, 2011). Other more proactive ap-proaches, such as using the NSP (Neighborhood Stabilization Pro-gram) and other federal, state and local initiatives, are provenhelpful in stabilizing and revitalizing neighborhoods heavilyaffected by foreclosures. Community Land Trust, “first look” ini-tiatives, encouraging real estate investment of foreclosed proper-ties, or simply strengthening code enforcement to pressure thefinancial institutions owing these REO properties to keep up withthe maintenance of the foreclosed properties are some othermeasures that local communities can rely on to reduce the effect offoreclosures on neighborhood disorder and dissatisfaction. Webelieve all these efforts would benefit the residents who survivedbut had to deal with the aftermath of the housing crisis.

This study demonstrates the utility of GIS technology to linkverbal responses with spatial data. Verbal responses, such as sur-veys and interviews, are the most widely used sources of data insocial science to learn about others' beliefs, feelings, or plans.However, linking them with spatial datadgeographically illus-trated datadhas great potential to lead to a better understanding ofspatial factors related to human responses (Hur, 2008). Thisresearch provides an exemplary example of this type of linkage bycombining subjective survey data and objective geographic infor-mation data to better understand the associations of both aspects.

By using the existing 2004 Homeowner Satisfaction Survey inour analysis, this research is subject to data limitation. Although thesurvey was comprehensive and it covered a variety of aspectsrelated to respondents' residential satisfaction, the survey excludedrenters from the sample, a limitationwhich could affect the validityof the research. Those neighborhoods with a high degree of fore-closure intensity in Franklin County (Fig. 2) are neighborhoods witha high percentage of renter-occupied housing and readers shouldbe aware of this limitation.

We point to several directions for future research. First, ourresearch methodology of using distance to foreclosed homes toexplain neighborhood satisfaction of homeowners will contributeto the literature. We have used four distance measures of fore-closure intensity: the ratio of foreclosed homes within 1/8-mile,1/4-mile, 1/2-mile, and 1-mile radii. Among them, the 1/4-mile radiusturned out to be the best representation of the Foreclosure intensityfactor in our model, followed by the 1/8-mile radius. The SEMmodel suggests excluding foreclosure ratio within 1/2-mile radiusand foreclosure ratio within 1-mile radius from the model. Weconclude that residents' perceptions have no association withforeclosure intensity in the areas further away. The geographic limitat which residents consider foreclosed homes to be problematic interms of residents' sense of community life is not clearly estab-lished, and future research may need to define this threshold. Sincedifferent neighborhoods and subdivisions have differentgeographical boundaries, sizes, shapes, and barriers, and sincepeople's perceptions of neighborhoods may differ based onneighborhood characteristics, case studies in specific neighbor-hoods or subdivisions will be very helpful in explaining morespecifically how foreclosures are linked to residential satisfaction.

Second, we understand that our foreclosure data reflects con-ditions prior to the subprime housing crisis. Even before thehousing foreclosure crisis, though, concentrated foreclosed housing

M. Hur et al. / Applied Geography 58 (2015) 7e1716

had negative effects on residents' neighborhood satisfaction. Sincethe impacts during the national recession were stronger, we as-sume the findings using data including the national recession datawill be much more interesting. By controlling other potential fac-tors that may affect neighborhood change and residents' neigh-borhood perceptions, a time series analysis of the relationshipcould show the dynamics of impacts of foreclosures at the neigh-borhood scale.

Third, we also assume that the neighborhood effects of fore-closures might be different in different markets. For example, re-sults in Florida where the housing foreclosure crisis was strongerand deeper may be different than results in Ohio. Comparisonsbetween markets may result in better insights on foreclosures indifferent markets, because even though the market dynamics aredifferent, the methodology and theory should hold.

Finally, although we now understand that the foreclosure in-tensity in a neighborhood negatively affects residents' perceptionsof neighborhood, we still do not know what kind of characteristicsreally provoke these perceptions and how they affect residents'perceptions and evaluations of a neighborhood. The characteristicscould be low maintenance efforts on the properties after fore-closure, the possibility of vandalism, residents' fear of foreclosureintensity in the neighborhood lowering their property values, thetype of neighborhood (city, inner suburb, and outer suburb), landuse mix, housing type, or all of these characteristics combined.Further research in these areas can help to understand how otherindividual and neighborhood physical, social, and economic factors,in addition to foreclosures, contribute to residents' neighborhoodsatisfaction or dissatisfaction.

References

Agarwal, S., Ambrose, B. W., Chomsisengphet, S., & Sanders, A. B. (2012). Thyneighbor's mortgage: does living in a subprime neighborhood affect one'sprobability of default? Real Estate Economics, 40(1), 1e22.

Alvi, S., Schwartz, M. D., DeKeseredy, W. S., & Maume, M. O. (2001). Women's fear ofcrime in Canadian public housing. Violence Against Women, 7(6), 638e661.

Amerigo, M., & Aragones, J. I. (1997). A theoretical and methodological approach tothe study of residential satisfaction. Journal of Environmental Psychology, 17(1),47e57.

Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: areview and recommended two-step approach. Psychological Bulletin, 103(3),411e423.

Arnio, A. N., Baumer, E. P., & Wolff, K. T. (2012). The contemporary foreclosure crisisand US crime rates. Social Science Research, 41(6), 1598e1614.

Batson, C. D., & Monnat, S. M. (2014). Distress in the desert: neighborhood disorder,resident satisfaction, and quality of life during the Las Vegas foreclosure crisis.Urban Affairs Review, 1e30. http://dx.doi.org/10.1177/1078087414527080.

Baumer, E. P., Arnio, A. N., & Wolff, K. T. (2013). Assessing the role of mortgage fraud,confluence, and spillover in the contemporary foreclosure crisis. Housing PolicyDebate, 23(2), 299e327.

Baxter, V., & Lauria, M. (2000). Residential mortgage foreclosure and neighborhoodchange. Housing Policy Debate, 11, 675e699.

Bennett, G. G., Scharoun-Lee, M., & Tucker-Seeley, R. (2009). Will the public's healthfall victim to the home foreclosure epidemic? PLoS Medicine, 6(6), e1000087.

Boardman, J. D. (2004). Stress and physical health: the role of neighborhoods asmediating and moderating mechanisms. Social Science & Medicine, 58(12),2473e2483.

Bocian, D. G., Ernst, K. S., & Li, W. (2008). Race, ethnicity and subprime home loanfinancing. Journal of Economics and Business, 60, 110e124.

Brown, B. B., Perkins, D. D., & Brown, G. (2004). Incivilities, place attachment andcrime: block and individual effects. Journal of Environmental Psychology, 24,359e371.

Bruin, M. J., & Cook, C. C. (1997). Understanding constraints and residential satis-faction among low-income single-parent families. Environment and Behavior,29(4), 532e553.

Carswell, A. T., James, R. N., & Mimura, Y. (2009). Examining the connection be-tween housing counseling practices and long-term housing and neighborhoodsatisfaction. Community Development. ISSN: 1557-5330, 40(1), 37e53.

CBS News 60 Minutes. (2011, Dec. 27). There goes the neighborhood. Retrieved20.07.14, from http://www.cbsnews.com/news/there-goes-the-neighborhood-57344513/.

Chan, S. W., Gedal, M., Been, V., & Haughwout, A. (2013). The role of neighborhoodcharacteristics in mortgage default risk: evidence from New York city. Journal ofHousing Economics, 22(2), 100e118.

Drassopoulos, A., Batson, C. D., Futrell, R., & Brents, B. G. (2012). Neighborhoodconnections, physical disorder, and neighborhood satisfaction in Las Vegas.Urban Affairs Review, 48(4), 571e600.

Ellen, I. G., Lacoe, J., & Sharygin, C. A. (2013). Do foreclosures cause crime? Journal ofUrban Economics, 74, 59e70.

Galster, G. (1987). Homeowners and neighborhood reinvestment. Durham, NC: DukeUniversity Press.

Gandel, S. (2011, Aug. 1). Bulldoze: the new way to foreclose. Time. Retrieved20.07.14, from http://business.time.com/2011/08/01/bulldoze-the-new-way-to-foreclose/.

Gramlich, E. M. (2007). Subprime mortgages: America's latest boom and bust.Washington, D.C.: Urban Institute Press.

Grinstein-Weiss, M., Yeo, Y., Anacker, K., van Zandt, S., Freeze, E. B., & Quercia, R. G.(2011). Homeownership and neighborhood satisfaction among low-and-moderate income households. Journal of Urban Affairs, 33(3), 247e265.

Grover, M., Smith, L., & Todd, R. M. (2008). Targeting foreclosure interventions: ananalysis of neighborhood characteristics associated with high foreclosure ratesin two Minnesota counties. Journal of Economics and Business, 60, 91e109.

Hair, J. F., Jr., Black, W. C., Babin, B. J., & Anderson, R. E. (2009). Multivariate dataanalysis (7th ed.). Upper Saddle River: Pearson.

Han, H.-S. (2014). The impact of abandoned properties on nearby property values.Housing Policy Debate, 24(2), 311e334.

Hatcher, L. (1994). A step-by-step approach to using the SAS system for factor analysisand structural equation modeling. Cary, North Carolina: SAS Institute Inc.

Hill, T. D., & Angel, R. J. (2005). Neighborhood disorder, psychological distress, andheavy drinking. Social Science & Medicine, 61(5), 965e975.

Hur, M. (2008). Neighborhood satisfaction, physical and perceived characteristics(Doctoral dissertation). Retrieved from ProQuest dissertations and theses.(Accession order no. 3338574).

Hur, M., & Morrow-Jones, H. (2008). Factors that influence residents' satisfactionwith neighborhoods. Environment & Behavior, 40(5), 619e635.

Hur, M., & Nasar, J. L. (2014). Physical upkeep, perceived upkeep, fear of crime, andneighborhood satisfaction. Journal of Environmental Psychology, 38, 186e194.

Immergluck, D., & Smith, G. (2005). Measuring the effect of subprime lending onneighborhood foreclosures: evidence from Chicago. Urban Affairs Review, 40(3),362e389.

Immergluck, D., & Smith, G. (2006a). The impact of single family mortgage fore-closures on neighborhood crime. Housing Studies, 21(6), 851e866.

Immergluck, D., & Smith, G. (2006b). The external costs of foreclosure: the impact ofsingle-family foreclosures on property values. Housing Policy Debate, 17(1),57e79.

Johnson, M. P., Turcotte, D. A., & Sullivan, F. M. (2010). What foreclosed homesshould a municipality purchase to stabilize vulnerable neighborhoods? Net-works & Spatial Economics, 10(3, SI), 363e388.

Johnson, T. P., O'Rourke, D., Burris, J., & Owens, L. (2002). Culture and surveynonresponse. In R. M. Groves, D. A. Dillman, J. L. Eltinge, & R. J. A. Little (Eds.),Survey nonresponse (pp. 55e69). John Wiley & Sons.

Kim, J. (2010). Neighborhood disadvantage and mental health: the role of neigh-borhood disorder and social relationships. Social Science Research, 39(2),260e271.

Kim, J., & Kaplan, R. (2004). Physical and psychological factors in sense of com-munity e new urbanist Kentlands and nearby Orchard Village. Environment andBehavior, 36(3), 313e340.

Kline, R. B. (2005). Principles and practice of structural equation modeling (2nd ed.).New York, London: The Guilford Press.

Latkin, C. A., & Curry, A. D. (2003). Stressful neighborhoods and depression: aprospective study of the impact of neighborhood disorder. Journal of Health &Social Behavior, 44(1), 34e44.

Leonard, T., & Murdoch, J. C. (2009). The neighborhood effects of foreclosure. Journalof Geographical Systems., 11(4), 317e332.

Lin, Z., Rosenblatt, E., & Yao, V. W. (2009). Spillover effects of foreclosures onneighborhood property values. Journal of Real Estate Finance and Economics, 38,387e407.

Long, D. A., & Perkins, D. D. (2003). Confirmatory factor analysis of the sense ofcommunity index and development of a brief SCI. Journal of Community Psy-chology, 31(3), 279e296.

Lucy, W. H. (2010). Foreclosing the dream. Chicago, IL, Washington, DC: APA PlannersPress.

Marans, R. W., & Rodgers, W. (1975). Toward an understanding of communitysatisfaction. In A. H. Hawley, & V. P. Rock (Eds.), Metropolitan America incontemporary perspective (pp. 299e352). New York: Sage Publications.

Miller, F. D., Tsemberis, S., Malia, G. P., & Grega, D. (1980). Neighborhood satisfactionamong urban dwellers. Journal of Social Issues, 36(3), 101e117.

National Bureau of Economic Research. (2014). U.S. business cycle expansions andcontractions. Retrieved 21.07.14, from http://www.nber.org/cycles.html.

O'Brien, D. T., & Wilson, D. S. (2011). Community perception: the ability to assess thesafety of unfamiliar neighborhoods and respond adaptively. Journal of Person-ality and Social Psychology, 100(4), 606e620.

Patterson, P. K., & Chapman, N. J. (2004). Urban form and older residents' serviceuse, walking, driving, quality of life, and neighborhood satisfaction. AmericanJournal of Health Promotion, 19(1), 45e52.

RealtyTrac. (2014). Retrieved 27.03.14, from http://RealtyTrac.com.Rohe, W. M., & Stegman, M. A. (1994). The effects of homeownership: on the self-

esteem, perceived control and life satisfaction of low-income people. Journalof the American Planning Association, 60(2), 173e184.

M. Hur et al. / Applied Geography 58 (2015) 7e17 17

Rohe, W. M., & Stewart, L. S. (1996). Homeownership and neighborhood stability.Housing Policy Debate, 7(1), 37e81.

Sampson, R. J., Raudenbush, S. W., & Earls, F. (1997). Neighborhood and vio-lent crime: a multilevel study of collective efficacy. Science, 277(15),918e924.

Schildkraut, J., & Mustaine, E. E. (2013). Movin', but not up to the east side: fore-closures and social disorganization in Orange County, Florida. Housing Studies,29(2), 177e197.

Schreiber, J. B., Stage, F. K., King, J., Nora, A., & Barlow, E. A. (2006). Reportingstructural equation modeling and confirmatory factor analysis results: a review.The Journal of Educational Research, 99(6), 323e337.

The Federal Reserve. (2006). Frequently asked questions about the new HMDA data.Retrieved 20.07.14, from http://www.federalreserve.gov/newsevents/press/bcreg/bcreg20060403a1.pdf.

Wallace, D., Hedberg, E. C., & Katz, C. M. (2012). The impact of foreclosures onneighborhood disorder before and during the housing crisis. Social ScienceQuarterly, 93(3), 625e647.

Webb, M. D. (2009). Subprime lending, the housing bubble, and foreclosures in Lima,Ohio (Master's thesis). The Ohio State University.

Weidemann, S., & Anderson, J. R. (1985). A conceptual framework for residentialsatisfaction. In I. Altman, & C. M. Werner (Eds.), Home environments (Vol. 8).New York & London: Plenum Press.

Westaway, M. S. (2007). Life and neighborhood satisfaction of black and whiteresidents in a middle-class suburb of Johannesburg. Psychological Reports,100(2), 489e494.

Zhang, H., & McCord, E. S. (2014). A spatial analysis of the impact of housingforeclosures on residential burglary. Applied Geography, 54, 27e34.