Can community land trusts slow gentrification?

Abstract

Gentrification has become one of the most widely discussed phenomena in urban affairs due to its prevalence around the world. However, most discussions remain at the theoretical level or are limited to case studies of a few metropolitan cities. Furthermore, there is little research on the relationship between gentrification and community land trusts (CLTs)—to our knowledge, no studies have examined their connection.
Our investigation seeks to address this research gap by evaluating whether and how CLTs affect gentrification. Can CLTs counteract the negative effects of gentrification? Using a data set detailing the locations of CLT units at the national level, we employed a binomial logistic regression to examine whether CLTs influence gentrification, as well as a paired t-test with 9 relevant indicators to understand how CLTs affect gentrification. The findings suggest that CLTs may function as appropriate tools for stabilizing neighborhoods at risk of gentrification.

Full text

Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=ujua20
Download by: [Korea Research Institute for Human Settlements] Date: 27 September 2017, At: 19:17
Journal of Urban Affairs
ISSN: 0735-2166 (Print) 1467-9906 (Online) Journal homepage: http://www.tandfonline.com/loi/ujua20
Can community land trusts slow gentrification?
Myungshik Choi, Shannon Van Zandt & David Matarrita-Cascante
To cite this article: Myungshik Choi, Shannon Van Zandt & David Matarrita-Cascante
(2017): Can community land trusts slow gentrification?, Journal of Urban Affairs, DOI:
10.1080/07352166.2017.1362318
To link to this article: http://dx.doi.org/10.1080/07352166.2017.1362318
Published online: 27 Sep 2017.
Submit your article to this journal
View related articles
View Crossmark data

Can community land trusts slow gentrification?
Myungshik Choi
a
, Shannon Van Zandt
b
, and David Matarrita-Cascante
b
a
Korea Research Institute for Human Settlements;
b
Texas A&M University
ABSTRACT
Gentrification has become one of the most widely discussed phenomena in
urban affairs due to its prevalence around the world. However, most
discussions remain at the theoretical level or are limited to case studies of
a few metropolitan cities. Furthermore, there is little research on the rela-
tionship between gentrification and community land trusts (CLTs)—to our
knowledge, no studies have examined their connection.
Our investigation seeks to address this research gap by evaluating
whether and how CLTs affect gentrification. Can CLTs counteract the nega-
tive effects of gentrification? Using a data set detailing the locations of CLT
units at the national level, we employed a binomial logistic regression to
examine whether CLTs influence gentrification, as well as a paired t-test
with 9 relevant indicators to understand how CLTs affect gentrification. The
findings suggest that CLTs may function as appropriate tools for stabilizing
neighborhoods at risk of gentrification.
Despite increasing concerns about gentrification in many of our nation’s metropolitan areas,
relatively few strategies have been identified that can mitigate its negative effects. Such strategies
are typically developed through valuable research. However, though gentrification has been widely
discussed, most conversations remain at the theoretical level or are limited to case studies of typical
American metropolitan cities. More specifically, studies examining the relationship between gentri-
fication and housing programs are rare—to our knowledge, no studies have empirically assessed the
relationship between gentrification and community land trusts (CLTs).
CLTs are perceived to offer benefits for neighborhoods, but few researchers have examined the
practical impact of CLTs on their communities. Even fewer studies have evaluated the ability of
CLTs to influence changes in neighborhood conditions such as gentrification. For this reason, our
research addresses this lack of empirical research by evaluating the effects of CLTs on gentrification.
More specifically, it poses the following research question: Do CLTs counteract the known negative
effects of gentrification on neighborhoods?
Gentrification and CLTs
Gentrification first gained attention within the planning field in the 1980s when discussions on its
causes resulted in fierce debates between supply-side and demand-side theorists (Bailey & Robertson,
1997).
1
Proponents of supply-side theories argued that gentrification, as an outcome of capitalism,
stemmed from flows of capital that favored developers, landlords, and investors (Smith, 1979; Smith
& LeFaivre, 1984). However, demand-side theorists claimed that such arguments failed to adequately
explain the causes of gentrification despite numerous empirical assessments and case studies
published between the 1970s and early 1980s (Beauregard, 1986; Ley, 1996). Clay (1979) argued
CONTACT Myungshik Choi mschoi@krihs.re.kr Korea Research Institute for Human Settlements, 5 Gukchaegyeonguwon-
ro, Sejong-si 30147, Republic of Korea.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/ujua.
© 2017 Urban Affairs Association
JOURNAL OF URBAN AFFAIRS
https://doi.org/10.1080/07352166.2017.1362318
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

from a demand-side perspective that gentrification resulted from the resettlement of professional
and upper-middle-class homeowners into city neighborhoods. This class of gentrifiers included
diverse groups of professionals, artists, and people wealthier than previous residents (Redfern,
2003). These demographics typically included nontraditional households, couples with later mar-
riages, families with fewer children, gay couples, childless marriages, nonmarried couples, and single
people (Lang, Hughes, & Danielsen, 1997). Thus, proponents of demand-side theories emphasized
demographic factors as more compelling causes of gentrification than flows of capital.
In the early 1990s, however, demand-side and supply-side theories transitioned from a conflicting
to a complementary relationship (Hamnett, 1991; Lees, Slater, & Wyly, 2000; Ley, 1996). Wyly and
Hammel (1999) observed the resurgence of gentrification after a mid-1990s economic recession and
attributed it not only to class turnover (demand-side perspective) but also to capital reinvestment
(supply-side perspective). Indeed, early supply-side and demand-side explanations still offer mean-
ingful contributions to the understanding of gentrification during this decade (Lees et al., 2008).
Perhaps because of these broad and shifting debates, gentrification has become perceived as a
chaotic and complex phenomenon. Because there is no unified theory of the gentrification process,
the term itself resists a singular definition (Beauregard, 1986; Davidson, 2011; Rose, 1984). Some
scholars regard gentrification as a positive process that encourages renovation, upgraded dwellings
(Rose, 1984), and reversals from decline or disinvestment in inner-city neighborhoods (Freeman,
2005). Others interpret gentrification as a negative phenomenon—indeed, the majority of scholars
write that the negative aspects outweigh the positives. For example, the definition of gentrification as
displacement of the working class by the upper-middle class emphasizes such relocation as the
primary negative consequence of this process (Kennedy & Leonard, 2001; Sumka, 1979; Wyly &
Hammel, 1999).
Considering its diverse dynamics worldwide and its multiple controversial definitions, gentrifica-
tion should be studied in terms of its actual phases and effects, rather than its definitions (Lees et al.,
2008). Gentrification includes not only physical or spatial restructuring but also dimensions of social,
cultural, political, economic, and institutional significance (Bourne, 1993; Lees et al., 2008; Smith &
LeFaivre, 1984; Smith & Williams, 1986; Zukin, 1987). For the purposes of this research, we
acknowledged that gentrification depends partly on capital reinvestment toward an urban core,
but we also examined factors that indicated resident class shifts in a neighborhood from working
class to upper-middle class, along with involuntary displacement of the existing working class.
The negative neighborhood effects associated with gentrification have caused major concerns.
Among them, three are central to the debate. The first is the displacement of incumbent residents.
Though some argue that displacement does not occur or is unrelated to gentrification (Freeman,
2005; Freeman & Braconi, 2004), the majority of researchers assert that gentrification leads directly
to displacement (Angotti, 2012; Clay, 1979; Smith, 1996).
Displacement as a corollary to gentrification forces longtime residents and businesses to relocate
due to skyrocketing land prices and rents (Angotti, 2012; Lees et al., 2008). The influx of investment,
ostensibly improving the environment, also brings with it the potential for destruction of the
community (Abu-Lughod, 1994; Betancur, 2002). According to Betancur (2011), the real tragedy
of gentrification includes not only displacement but also community disintegration. Angotti (2012)
claimed, “Gentrification is not place-making but place-taking”(p. 103). Other criticisms cite the
disappearance of small businesses, changes in established neighborhood identities (Brown-Saracino,
2004), and disruption of social dynamics. Critics have concluded that such drawbacks characterize
gentrification efforts as poor strategies for urban revitalization (Angotti, 2012; Betancur, 2002).
Second, gentrification often leads to the removal of affordable housing from urban building stock
(Zukin, 1987). Along with displacement, the decrease in affordable units diminishes housing
opportunities for low-income households in gentrified neighborhoods. Such housing units are
crucial for many low-income, immigrant, and minority communities who experience a shortage of
options concerning where to live (Betancur, 2002).
2M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Unsustainable speculative property price increases comprise the third negative effect of gentrifi-
cation. Poor people have fewer resources and less power to enact institutional resistance in neigh-
borhoods with a speculative boom. Consequently, all neighborhoods are subject to the potential
threat of the rent increase process (Logan & Molotch, 1987). Even advocates of gentrification point
to the inflation of housing prices as a chief drawback associated with gentrification (Freeman, 2005).
Many policy tools and programs implemented by governments, communities, and/or combined
efforts have mitigated gentrification and its effects. These initiatives are categorized into two
strategies: improvement of housing affordability and the building of community assets (Levy,
Comey, & Padilla, 2007; Wyly & Hammel, 1999). However, these strategies—providing one-time
affordable housing units and not separating land from the speculative market—remain insufficient.
New mechanisms such as CLTs are emerging to compensate for this lack of policies. The CLT is a
relatively new housing model in the United States, mitigating the negative effects of gentrification by
offering long-term owner-occupied affordable housing and building community assets for low-
income households (Davis, 2010; Defilippis, 2004; Lees, 2008). Since the emergence of the first
CLT (New Communities Inc., founded in 1969), their numbers have increased to over 240 in 45
states. Though most CLTs are nonprofit organizations, state and city governments have increasingly
implemented CLTs as components of broader housing programs. In addition, other countries have
adopted the CLT model and concept (Davis, 2010).
CLTs assume that land is a public asset and not a private good. Under this principle, a community
—often a nonprofit organization—owns and leases land to individual residents who buy structures
on the land from CLTs. Typically, CLTs acquire land through donation or purchase, offering it
through long-term ground lease to residents who own their homes. CLTs impose restrictions on the
resale prices of their units to preserve affordability, and they serve as long-term stewards of the land
(Curtin & Bocarsly, 2008; Davis, 2007; Gray, 2008). Their membership follows an open, place-based
system, with the board of directors including residents of CLT units, other community residents, and
public representatives, to allow for a self-governing community.
The primary purpose and benefit of CLTs is to preserve long-term affordable housing in
neighborhoods by removing houses, buildings, and lands from the market (Curtin & Bocarsly,
2008;Gray,2008; Paterson & Dunn, 2009). Secondly, CLTs enhance neighborhood stability by
increasing length of residency, preventing displacement of low-income households, and main-
taining optimal unit conditions by ensuring security of tenure. Moreover, CLTs can prevent
excessive speculative investment from gentrifying areas by preserving housing affordability
(Davis, 1991; Saegert & Benitez, 2005). Thirdly, CLTs contribute to the building of community
assets, because owner-occupants interact regularly on the basis of shared residential interests. In
addition, CLTs promote increases in economic and racial diversity by creating and preserving
affordable housing units in neighborhoods where low-income families would not otherwise be
able to live (Davis, 2006).
These three benefits of CLTs can mitigate the major negative effects of gentrification. By building
community assets, producing and preserving affordable housing, and stabilizing neighborhoods,
CLTs might prevent the displacement of low-income households, in addition to counteracting low
affordability and rising property values in neighborhoods.
Research framework
The central question of this study addresses whether and how CLTs counteract the negative effects of
gentrification. Our analytical approach involves a quasi-experimental, pretest–posttest, nonequiva-
lent groups design. Four neighborhood groups were established in order to assess their changes from
2000 to 2010. The impact of CLTs (the treatment) on gentrifying and nongentrifying neighborhoods
was tested, while control groups (both gentrifying and nongentrifying neighborhoods without CLTs)
were maintained, as seen in Figure 1.
JOURNAL OF URBAN AFFAIRS 3
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Figure 1 describes how the four groups of neighborhoods were categorized and named through-
out the remainder of the article: nongentrifying neighborhoods with CLTs (nG/C), nongentrifying
neighborhoods without CLTs (nG/nC), gentrifying neighborhoods with CLTs (G/C), and gentrifying
neighborhoods without CLTs (nG/nC).
Based on our literature review, we hypothesize that CLTs have the potential to offset gentrifica-
tion in three ways: (a) counteracting displacement, (b) preserving affordability, and (c) stabilizing the
speculative increase of property values when neighborhoods are gentrified. Nine indicators
(described in Table 1) were selected based on these three categories to measure whether and how
CLTs affect gentrification. We employed binomial logistic regression to assess whether CLTs have an
independent impact on gentrification, after which we measured how CLTs affect gentrification
through paired t-tests. Comparisons across groups allowed us to examine the impacts of CLTs on
neighborhoods that were gentrifying and those that were not.
Unit of analysis
The census tract is the most commonly used proxy for neighborhoods in the literature on neighborhood
effects and/or gentrification (Freeman, 2005;Galster&Booza,2007; Hammel & Wyly, 1996), though
the literature acknowledges the ambiguity of neighborhood boundaries (Chaskin, 1997; Park & Rogers,
Table 1. Neighborhood indicators for measurement (data from the 2000 and 2010 decennial censuses).
Indicator Description
Racial composition Proportion of White population
Lower-middle class Proportion of households with 80–100% of AMI
Income level Proportion of median income compared to citywide area
Education level Proportion of residents who graduated high school or more
Length of residence Proportion of residents who live in the same house compared to citywide area
Age fluctuation
a
Change in age distribution between 2000 and 2010 (index)
Affordability Proportion of median housing value compared to citywide area
Homeownership Proportion of owner-occupied housing units compared to all occupied housing units
Housing value Proportion of median housing value compared to national median housing value
Note.
a
We composed the age fluctuation index, which indicates the amount of change in age distribution between 2000 and 2010.
There are four age brackets in the census data set. Each bracket has its own percentage in a neighborhood, so the age
fluctuation index is calculated by the sum of the percentage change in each bracket between 2000 and 2010 according to the
following table and formula:
Description of the age fluctuation index formula
Age bracket 2000 2010 Change
Under 18 years A0 A1 (A1 −A0)
2
18 to 34 years B0 B1 (B1 −B0)
2
35 to 64 years C0 C1 (C1 −C0)
2
65 and over D0 D1 (D1 −D0)
2
Age Fluctuation Index Formula ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
A1 A0ðÞ^2þB1 B0ðÞ^2þC1 C0ðÞ^2þD1 D0ðÞ^2
p.
Nongentrifying (nG/C)
Nongentrifying (nG/nC)
Gentrifying (G/C)
Gentrifying (G/nC)
2000 CLTs 2010
nG X nG
nG nG
GXG
GG
Figure 1. Research design.
4M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

2015). On the other hand, neighborhoods are generally considered to be smaller than cities but larger
than a small collection of blocks (Sawicki & Flynn, 1996). According to the U.S. Census Bureau, census
tracts generally contain 4,000 people, never cross any state or county boundaries, and include relatively
homogeneous units grouped according to population characteristics, economic status, and living
conditions. In this sense, census tracts roughly approximate the typical concept of neighborhoods.
Therefore, consistent with the existing literature, this research employs census tracts as units of analysis,
though it uses the term neighborhood in lieu of census tract.
Sample selection and measurement approach
Data collection
To assess the impact of CLTs on neighborhoods, we first must establish the number, location, and
establishment dates of CLT units. To collect the location and number of units, each CLT organiza-
tion listed in the directory of the National Community Land Trust network was contacted during the
summer of 2014. Of the 131 baseline CLTs that were contacted, a total of 68 responded, reflecting a
response rate of 52%. However, only 46 provided the requested information, including the locations
of 3,709 CLT units in 22 states. The remaining 22 CLTs revealed that they had not yet established
any CLT units on the ground.
Although we expected units from the same CLT organizations to be clustered, this was not always
the case. To ensure that we examined a “critical mass”of units, we excluded neighborhoods that had
only one or two CLT units. In this study, only neighborhoods with three or more units were
considered neighborhoods with a CLT.
2
To accurately assess the impact of CLTs, we used only
CLT units that were put into service between 2000 and 2009, allowing us to include nearly 80%
(79.5%) of all CLT units currently in service. Upon implementing these criteria, we were left with
124 neighborhoods with three or more CLT units.
Measurement of gentrification
Given our comparison of neighborhoods categorized based on whether they were gentrifying, the
identification of gentrifying neighborhoods was critical. We relied on previous literature to identify
indicators appropriate for determining whether a neighborhood was gentrifying (Atkinson, 2000;
Freeman, 2005; Freeman & Braconi, 2004; Galster & Peacock, 1986; Hammel & Wyly, 1996; Sullivan,
2007). Using data from the U.S. Census and the American Community Survey, changes in the
following five characteristics were used to identify gentrifying neighborhoods: racial composition,
education level, neighborhood average income, property values, and homeownership types.
3
To
minimize unexpected effects from diverse time spans and to reflect recent conditions, the years 2000
and 2010 were selected to exhibit the rate of change in these five characteristics within each
neighborhood. To be identified as a gentrifying area, neighborhoods were required to simultaneously
meet all of the following quantitative criteria:
(1) The rate of change of percentage White population is higher than that of the correspond-
ing citywide median.
4
(2) The rate of change of percentage college-educated is higher than that of the corresponding
citywide median.
(3) The rate of change of median income is higher than 120%
5
of the corresponding citywide
median.
(4) The rate of change of median value of single-family homes is higher than that of the
corresponding citywide median.
(5) The rate of change of percentage owner-occupied units is higher than that of the corre-
sponding citywide median.
JOURNAL OF URBAN AFFAIRS 5
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

In summary, the changes over time and relative values compared to each neighborhood’s corre-
sponding citywide area were investigated as the primary criteria by which to identify gentrifying
neighborhoods.
Selection of each neighborhood type
Out of the 124 neighborhoods with CLTs, 14 neighborhoods in nine cities and five states were
deemed to be gentrifying between 2000 and 2010 (G/C); the remaining 110 neighborhoods in 15
states were regarded as nongentrifying neighborhoods with CLTs (nG/C).
To identify the comparison groups without CLTs (G/nC and nG/nC), we chose neighborhoods
located adjacent to or near the neighborhoods with CLTs in order to control for larger socio-
economic differences. To select G/nC neighborhoods, every neighborhood in the nine cities with G/
C neighborhoods was investigated for evidence of gentrification. Of the 388 neighborhoods, 38 were
deemed to be gentrifying and comprised our comparison group (G/nC). We added six more G/nC
neighborhoods by considering tracts adjacent to CLT tracts in cities without G/C neighborhoods, for
a total of 44 G/nC neighborhoods.
To select the remaining nongentrifying non-CLT neighborhoods (nG/nC) from the process
identified above, we selected 52 nongentrifying neighborhoods adjacent to neighborhoods with
CLTs. Afterward, we randomly selected 52 neighborhoods from the remaining pool of 469 tracts
in our CLT cities that did not have CLTs and were not gentrifying, resulting in a total of 104 tracts as
shown in Table 2. Altogether, our four comparison groups included 272 neighborhoods.
There remained a few limitations in the selection of neighborhoods. First, the limited number of
cities in the sample selection process renders difficult the generalization of the results. Second, a few
major CLT organizations in megacity regions were excluded from the analysis due to nonresponse.
Measurement of neighborhood impact
Several neighborhood indicators related to the drawbacks of gentrification and the benefits of CLTs
were selected based on literature review. First, in response to the threat of displacement, CLTs build
community assets in their neighborhoods, thereby increasing racial diversity, maintaining lower-
middle-class households and income levels, and stabilizing education levels. They also stabilize their
neighborhoods by increasing the length of residence and reducing age fluctuation.
6
Second, CLTs
counteract decreasing affordability in neighborhoods by instead prioritizing affordability and
decreasing owner-occupied housing rates. Third, they mitigate skyrocketing housing prices in their
neighborhoods by slowing the rise of housing prices compared to those of surrounding areas.
Table 1 describes how each indicator was measured.
Displacement transforms the demographic composition of neighborhoods. The research hypoth-
esis predicts that CLTs help maintain racial diversity in gentrifying neighborhoods. The proportion
of White residents to the total population often measures racial diversity in neighborhoods.
The low-income household ratio in a community indicates the level of a community’s assets, and
the building of community assets relates to the mixed-income residents in that community. Because
CLTs attract higher numbers of lower-middle-class households into their units, the research hypoth-
esis argues that CLTs maintain a middle-class ratio in neighborhoods that are gentrifying. The
definition of middle-class may vary, but this study focuses on the lower-middle class—the primary
target of CLTs. Generally, the lower-middle class accounts for 80 to 100% of the area median income
Table 2. Number of neighborhoods by type.
With CLT Without CLT
Gentrifying 14 (G/C) 44 (G/nC)
Nongentrifying 110 (nG/C) 104 (nG/nC)
6M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

(AMI). Therefore, the lower-middle-class ratio corresponds to the percentage of households with
incomes of 80 to 100% of the AMI.
In general, resident incomes indicate the assets available to a community (Kretzmann &
McKnight, 1993). However, when displacement occurs due to the influx of wealthier people
into areas that are gentrifying, the stabilization of income level may maintain existing community
assets by preventing displacement of low-income residents from such areas. In this sense, we
assume that CLTs stabilize neighborhood income levels in neighborhoods that are gentrifying.
The proportion of median income to citywide area can be used to estimate neighborhood income
level.
CLTs may stabilize neighborhoods located in rapidly changing areas. We hypothesize that less
change in stability indicators for neighborhoods (education level, length of residence, and age
fluctuation) than for citywide areas corresponds to lower levels of displacement for incumbent
residents.
CLTs strive primarily to maintain affordability by providing affordable housing units to middle-
and low-income households and expanding the number of permanently affordable dwelling units
within neighborhoods. Therefore, understanding how CLTs affect their neighborhoods in terms of
affordability proves critical to evaluate their effectiveness. We hypothesize that CLTs work to
counteract decreasing affordability in gentrifying neighborhoods. Although the housing affordability
may be defined a variety of ways, we defined the housing value of a neighborhood in comparison to
that of its surrounding area. Generally, housing values in gentrifying areas closely follow increases in
median income. This periodical mismatch between median income and housing value explains why
we refrain from using income–housing value ratios as an affordability index. Thus, the affordability
index is defined as the proportion of a neighborhood’s median housing value to that of a citywide
area.
7
CLTs aim to increase homeownership as a means of building community assets by leasing land
and selling houses to residents. In their neighborhoods, CLTs provide more housing options for
residents and, in turn, promote increases in the number of available rental units. Generally, rental
units are crucial for low-income residents and thus it is important to ensure affordable rental
unitsaslandpricesrise(Levyetal.,2007). The research hypothesis predicts that, especially in
gentrifying neighborhoods, CLTs help stabilize increasing owner-occupied housing rates, which
are defined as the percentage of owner-occupied housing units out of the total number of
occupied housing units.
Skyrocketing property prices in gentrifying neighborhoods serve as major negative effects of
gentrification that have been explored by researchers who focus on rent gap theory (Lees et al.,
2008;Smith,1996). When gentrification occurs, wealthier people move in and raise rents for
both residential and commercial use, displacing low-income people. Because rents and property
values are closely related, we hypothesize that CLTs can help alleviate abrupt property value
increases in neighborhoods that are gentrifying. In this sense, housing price acts as one of the
most common indicators of overall property values in neighborhoods. In addition, the stabiliza-
tion of housing price relates to the ability of CLTs to stabilize neighborhoods. We used the
housing price index—that is, the relative housing price compared to the national median housing
price—to compare changes between 2000 and 2010 without inflation effects. National median
housing prices were $111,800 in 2000 and $188,400 in 2010. Thus, the housing price index
indicates the proportion of a neighborhood’s median housing price to the national median
housing price.
Findings
This study explored whether and how CLTs affect gentrification. A binomial logistic regression
model was employed to examine whether CLTs yielded an impact on the phenomenon, and a paired
t-test was used to analyze how this impact was made.
JOURNAL OF URBAN AFFAIRS 7
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Logistic regression analysis
The logistic regression model explored the relationship between gentrification and a variety of
neighborhood indicators such as the existence of CLT units, while simultaneously controlling for
other indicators. The dependent variable measured whether the gentrifying process occurred in a
neighborhood between 2000 and 2010, and the independent variables included nine indicators from
the group comparison in 2010 as well as the existence of a CLT (dummy variable). The housing value
was not included as an independent variable due to high collinearity.
Multicollinearity
The possibility of multicollinearity persists as a common problem in most regression models. None
of the independent variables may be strongly correlated with another independent variable or with a
linear combination of another independent variable. To test for multicollinearity, we calculated
variance inflation factors (VIFs) and tolerance scores in Table 3.
All tolerance scores were greater than 0.2 and all VIFs were less than 4, confirming that there was
no issue with multicollinearity in our model. Generally, tolerance scores less than 0.2 or 0.1 and/or
VIFs greater than 5 or 10 indicate the presence of strong multicollinearity (Kutner, Nachtsheim, &
Neter, 2004;O’Brien, 2007).
Propensity score matching
Because our data were selected in nonexperimental settings, this study employed propensity score
matching to assign appropriate comparison groups. Propensity score matching is a tool used to
address confounding due to selection bias, reducing bias by accounting for selection effects
(Grinstein-Weiss et al., 2011). Propensity scores are the conditional probability that any neighbor-
hood in our studied area has CLT units, given a set of observed characteristics for the neighborhood
(Austin, 2011). The nearest neighbor matching without replacement was employed to select non-
CLT neighborhoods whose propensity scores were close to that of a CLT neighborhood. After
matching, the number of observations in our model decreased from 272 to 197 (Table 4).
Results
We developed three logistic regression models sequentially to disentangle the effects of CLTs and
other indicators. Model 1 includes demographic indicators, model 2 adds housing and class
indicators, and model 3 contributes the CLT indicator. Additionally, by including methodological
issues such as selection bias, we explored the relationship between gentrification and CLTs through
propensity score matching to reduce bias in model 3. We also presented the results of the logistic
regression analysis using matched samples.
Table 4 describes the results of the logistic regression from these three full sample models, as well
as from the matched sample model. The total number of neighborhoods was 272 for three full
sample models and 197 for the matched sample model.
Table 3. Multicollinearity diagnostics.
Variables VIF
SQRT
VIF Tolerance R
2
White population (%) 1.84 1.36 0.5425 0.4575
Graduate high school or more (%) 1.73 1.31 0.5786 0.4214
Age fluctuation between 2000 and 2010 1.06 1.03 0.9423 0.0577
Owner-occupied unit (%) 2.81 1.68 0.3559 0.6441
Length of residence compared to citywide area 1.99 1.41 0.5028 0.4972
Affordability index 2.38 1.54 0.4205 0.5795
Income level compared to citywide area 3.79 1.95 0.2638 0.7362
Lower-middle-class household (%) 1.13 1.06 0.8866 0.1134
CLT = 1 1.07 1.04 0.9307 0.0693
Mean VIF 1.98
8M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

As seen in Table 4, the presence of CLT units had a significant effect on the probability of
gentrification. We distinguished the effects of CLTs from those of other indicators by comparing
models.
Considering model 1, which was restricted to demographic characteristics, the findings con-
formed closely to the literature on gentrification. All other indicators were equal: for every 1%
increase in the age change index, the odds of gentrification increased by 15%. Moreover, for every
1% increase in the proportion of residents who graduated high school or achieved a higher level of
education, the odds of gentrification increased by 6%.
Model 2 included housing and class indicators and showed no significant findings with the
exception of age fluctuation, demonstrating that housing and class indicators do not substantially
affect gentrification.
Most important, model 3 illuminated the effects of CLTs on gentrification, adding to the analysis
with propensity score matched samples. The results of both models, as well as a CLT indicator,
provided strong support for our main hypothesis that predicted that CLTs might slow the gentrifica-
tion process. In the full sample model, when all other factors were equal, the odds of gentrification
were 0.30 times as likely for neighborhoods with CLTs than for neighborhoods without a CLT. In
other words, the presence of CLTs decreased the odds of gentrification by 70%. For the matched
sample model, when all other factors were equal, the odds of gentrification were 0.26 times as likely
for neighborhoods with CLTs than for neighborhoods without a CLT. In other words, the presence
of CLTs decreased the odds of gentrification by 74%, remaining significant beyond the 0.1% level.
The only difference between the full and matched sample models was the change in statistical
significance for the White population ratio. Unlike in the full sample, no significant relationship
existed between the White population and gentrification in the matched sample.
Although some differences existed, after propensity score matching was implemented, CLTs still
maintained a significant counteractive relationship with gentrification. Throughout the models, age
fluctuation consistently displayed a significant negative relationship with gentrification, as expected.
Goodness of fit
Assessing the fit between the estimated logistic regression model and the actual data set is crucial,
because it shows whether the model includes every important variable with the correct functional
form (Hosmer, Lemeshow, & Sturdivant, 2013). Though there is no simple goodness-of-fit measure
for a logistic regression model (Browne & Tootell, 1995), the Hosmer-Lemeshow test has been
commonly adopted, grouping cases based on the values of estimated probabilities. It groups data by
ordering the estimated probabilities and identifying 10 equally sized subgroups. Afterward, it
Table 4. Odds ratios of the likelihood of gentrification.
Variables
Model 1:
Demographic
Model 2: Housing and
class
Model 3: CLTs
Full
sample
Matched
sample
White population (%) 0.98 0.98 0.97* 0.98
Graduate high school or more (%) 1.06* 1.04 1.05 1.02
Age fluctuation between 2000 and 2010 1.15** 1.16** 1.15** 1.14**
Owner-occupied unit (%) 0.96 1.00 0.99
Length of residence compared to citywide area 1.90 1.17 1.89
Affordability index 1.04 0.95 0.72
Income level compared to citywide area 2.09 1.66 6.15
Lower-middle-class household (%) 0.94 0.95 0.98
CLT = 1 0.30*** 0.26***
Constant 0.00*** 0.00* 0.01 0.02
Log likelihood −132.07 −129.23 −123.27 −82.61
Number of neighborhoods 272 272 272 197
Note.*p< .05. **p< .01. ***p< .001.
JOURNAL OF URBAN AFFAIRS 9
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

calculates the observed and expected frequencies of gentrification and nongentrification for each
group. We estimated the goodness-of-fit test for the matched sample below.
The Hosmer-Lemeshow goodness-of-fit statistic calculated from the frequencies in Table 5 was
8.97, and the corresponding p-value calculated from the chi-square distribution with 8 degrees of
freedom was .3449. This suggests that we cannot reject our model, because it fits with high
agreement between the observed and expected cell frequencies.
The paired t-test
To explore how CLTs affect gentrification, we used a paired t-test to examine differences among
each of our control and treatment groups, both pre-CLT and post-CLT. The results are shown in
Tables 6–9. Nine indicators were tested based on the four neighborhood types in 2000 and 2010.
The differences between group means that were statistically significant at a level of p<.05were
considered to be valid, and Table 6 shows the results of each comparison.
Table 6. Comparison of CLT/non-CLT and gentrifying/nongentrifying neighborhoods over time by selected indicators.
Racial composition Lower-middle-class ratio Income level
2000 2010 Change 2000 2010 Change 2000 2010 Change
G/C 64.21 69.54 +5.33* 12.05 9.94 −2.11 0.98 1.10 +0.12*
nG/C 77.88 75.60 −2.28* 13.14 10.58 −2.56* 0.99 1.00 +0.01
G/nC 69.92 77.48 +7.56* 13.06 8.65 −4.41* 0.96 1.16 +0.20*
nG/nC 77.23 76.74 −0.49 12.44 9.68 −2.76* 1.09 1.08 −0.01
Education level Length of residence Age fluctuation
2000 2010 Change 2000 2010 Change Index
G/C 79.11 86.44 +7.33* 1.10 1.06 −0.04 6.59
nG/C 84.30 86.67 +2.37* 0.96 1.00 +0.04* 5.79
G/nC 81.55 91.49 +9.95* 0.90 1.01 +0.11* 7.73
nG/nC 83.34 86.79 +3.45* 1.02 1.01 −0.01 5.91
Affordability Homeownership Housing value
2000 2010 Change 2000 2010 Change 2000 2010 Change
G/C −0.91 −1.00 −0.09* 67.89 67.88 −0.01 0.90 1.12 +0.22*
nG/C −1.00 −1.00 0 62.92 60.70 −2.22* 1.31 1.43 +0.12*
G/nC −1.03 −1.20 −0.17* 56.88 58.46 +1.58* 1.33 1.53 +0.20*
nG/nC −1.10 −1.10 0 60.05 57.66 −2.39* 1.35 1.38 +0.03
Note. *p < .05.
Table 5. Goodness-of-fit (matched sample)—Observed and estimated expected frequencies within each
decile.
Gentrification = 1 Gentrification = 0
Decile Cut point Obs Exp Obs Exp Total
1 0.0514 0 0.8 20 19.2 20
2 0.0662 0 1.2 20 18.8 20
3 0.0841 3 1.5 17 18.5 20
4 0.1153 4 1.8 15 17.2 19
5 0.1511 3 2.7 17 17.3 20
6 0.1761 2 3.3 18 16.7 20
7 0.2329 2 4.0 17 15.0 19
8 0.2882 6 5.2 14 14.8 20
9 0.3836 6 6.6 14 13.4 20
10 0.7085 11 10.0 8 9.0 19
Note. Number of observations = 197. Number of groups = 10. Hosmer-Lemeshow chi
2
(8) = 8.97. Prob >
chi
2
= 0.3449.
10 M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Table 6 shows the overall results of the comparisons and the statistically significant differences
over time in each type of neighborhood. Both significant and insignificant differences were crucial in
the interpretation of differences among each test group. In Table 7, for example, significant t-values
in the rows indicate a significant change over time, whereas significant t-values in the columns
indicate differences between gentrifying and nongentrifying neighborhoods. Detailed interpretations
are provided by each indicator in the following tables.
To supplement the findings in Table 6,Table 7 compares gentrifying and nongentrifying
neighborhoods based on the three indicators that were not used for identifying gentrifying neigh-
borhoods: length of residence, affordability, and age fluctuation. Length of residence significantly
increased and affordability significantly decreased in gentrifying neighborhoods between 2000 and
2010, while showing no change in nongentrifying neighborhoods. Age fluctuation between 2000 and
2010 was significantly higher in gentrifying neighborhoods.
Table 8 shows the results of comparisons between neighborhoods with and without CLTs to show
the impact of CLTs on average neighborhoods (regardless of gentrification). Only three comparisons
(racial composition, affordability, and income level) showed statistically significant differences: The
proportion of the White population decreased in neighborhoods with CLTs and increased in those
without a CLT, meaning that racial diversity increased in neighborhoods with a CLT. However, the
affordability index and income level significantly decreased in neighborhoods without a CLT,
whereas no such change was evident in neighborhoods with CLTs.
The cross-comparison tables complement the findings in Table 6 and fully compare the means of
each neighborhood type. Even though comparisons were conducted for every indicator, only the
comparisons with statistically significant differences by year among the nine indicators are shown.
Racial composition
According to Table 6, significant increases in the White population are shown in both gentrifying
neighborhoods. By contrast, significant decreases in the White population of nongentrifying
Table 7. Comparisons between gentrifying and nongentrifying neighborhoods by selected indicators.
Length of residence Affordability Age fluctuation
2000 2010 t-Value 2000 2010 t-Value 2000–2010
Gentrifying 0.95 1.02 −2.04* −1.00 −1.16 −8.58* 7.45
Nongentrifying 0.99 1.01 −1.11 −1.05 −1.05 −0.17 5.85
t-Value 0.93 −0.74 0.95 −1.83 −3.48*
Note.*p< .05.
Table 8. Comparisons between neighborhoods with CLT and no CLT by selected indicators.
Racial composition Income level Affordability
2000 2010 t-Value 2000 2010 t-Value 2000 2010 t-Value
CLT 76.30 74.90 2.38* 0.99 1.01 −1.37 −1.00 −1.00 −0.92
No CLT 75.10 76.96 −2.84* 1.05 1.10 −3.19* −1.08 −1.13 −3.15*
t-Value −0.44 0.88 1.39 1.99* 2.05* 2.73*
Note.*p< .05.
Table 9. Cross-comparisons between each neighborhood type by year.
Education level (2010) Affordability (2010) Housing value (2000)
CLT No t-Value CLT No t-Value CLT No t-Value
Gentrifying 86.44 91.49 −2.03* −1.00 −1.20 −3.00* 0.90 1.33 −3.93*
Nongentrifying 86.67 86.79 −0.06 −1.00 −1.10 −1.72 1.31 1.35 −0.36
t-Value −0.05 2.48* 0.02 1.43 −2.29* −0.14
Note.*p< .05.
JOURNAL OF URBAN AFFAIRS 11
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

neighborhoods with CLTs (with no corresponding change in nongentrifying neighborhoods without
CLTs) indicate that CLTs yield positive effects on racial diversity in nongentrifying neighborhoods.
The fact that the White population ratio has remained unchanged in nongentrifying neighborhoods
without CLTs between 2000 and 2010 further supports this assertion.
According to Table 8, the White population ratio significantly decreased in neighborhoods with
CLTs while significantly increasing in neighborhoods without CLTs. This shows that CLTs generally
have positive effects on the racial diversity of their neighborhoods, regardless of gentrification.
Lower-middle-class ratio
According to Table 6, the lower-middle-class ratios of gentrifying neighborhoods with CLTs did not
significantly change between 2000 and 2010. However, there were significant decreases in the
neighborhoods without CLTs that were gentrifying, as well as in both nongentrifying neighborhoods.
In neighborhoods that are gentrifying, therefore, CLTs might have significant impact on maintaining
lower-middle-class ratios, suggesting a contribution to stabilization.
In summary, CLTs counteract the decrease of lower-middle-class residents in gentrifying neigh-
borhoods, possibly reducing the displacement of such residents due to gentrification. However, CLTs
do not affect the maintenance of lower-middle-class ratios in nongentrifying neighborhoods.
Income level
In Table 6, as we expected, gentrification increased the median income of neighborhoods compared
to citywide areas. Table 8 shows that whereas income level increased in neighborhoods without
CLTs, no such changes occurred in neighborhoods with CLTs. Therefore, despite a lukewarm impact
on the trends of gentrifying neighborhoods, CLTs help stabilize income level in neighborhoods,
perhaps leading to lower levels of displacement for low-income households.
In summary, although gentrification raises the income levels of neighborhoods by displacing low-
income households, CLTs mitigate this effect in neighborhoods that are gentrifying.
Education level
Though Table 6shows no difference between neighborhood types, Table 9 shows that, among
gentrifying neighborhoods, those without CLTs had higher increases of high school–educated
residents than did those with CLTs between 2000 and 2010. Therefore, CLTs may dampen the
increase in number of households with higher levels of education during the gentrification
process.
In summary, CLTs may help prevent the displacement of less-educated people and lower-income
households. Though CLTs may have negative effects toward building community assets such as
higher education levels, they may positively maintain a sense of belonging by preventing the
displacement of incumbent residents. However, CLTs do not appear to significantly affect the
maintenance of education level in nongentrifying neighborhoods.
Length of residence
According to Table 6, length of residence does not change in gentrifying neighborhoods with CLTs
or in nongentrifying neighborhoods without CLTs, but it increases in gentrifying neighborhoods
without CLTs and nongentrifying neighborhoods with CLTs. Table 7 shows that, when considering
neighborhoods regardless of whether they have CLTs, length of residence increases in gentrifying
neighborhoods while remaining unchanged in nongentrifying neighborhoods.
Considering together the above findings, CLTs might increase length of residence in nongentrify-
ing neighborhoods while having no such effect in gentrifying neighborhoods. This result is incon-
sistent with our expectation. In addition, contrary to the general assumption, gentrification increases
length of residence in a neighborhood.
12 M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Age fluctuation
Table 7 shows that age distribution changed more within gentrifying neighborhoods than in
nongentrifying neighborhoods. Typically, age fluctuation would be expected among gentrifying
neighborhoods because people move or become displaced in such neighborhoods. However, we
found no significant change based on the existence of CLTs in neighborhoods.
Affordability
Table 6 shows that affordability decreases in gentrifying neighborhoods, whereas no such change is
reported in nongentrifying neighborhoods. Tables 7 and 8show that gentrification has negative
effects on affordability, whereas CLTs stabilize these decreases in affordability. Moreover, afford-
ability significantly decreased in neighborhoods without a CLT.
Table 9 shows that, among gentrifying neighborhoods, affordability decreased more in neighbor-
hoods without a CLT between 2000 and 2010, indicating that CLTs stabilize the decreases in
affordability within gentrifying neighborhoods.
The results coincide with what was expected. In gentrifying neighborhoods, though neighbor-
hoods with a CLT showed slight decrease in affordability, neighborhoods without CLTs experienced
a more drastic decrease. Thus, CLTs mediate the decrease of affordability in gentrifying neighbor-
hoods, suggesting that CLTs can counteract the trend of lower affordability in gentrifying
neighborhoods.
Homeownership
According to Table 6, in gentrifying neighborhoods, whereas neighborhoods with CLTs show no
change in owner-occupied housing rates, neighborhoods without CLTs experience increases in these
rates. This suggests that rental housing units, which are typically more affordable residential options,
either increased in proportion or did not decrease in gentrifying neighborhoods. By contrast, both
nongentrifying neighborhoods experienced reductions in owner-occupied housing rates.
In summary, CLTs may help retain rental units in gentrifying neighborhoods, thus preventing the
loss of rental housing units by providing additional residential options.
Housing value
Tables 6 and 9show significant increases in housing price index in both gentrifying neighborhoods.
The housing price index of gentrifying neighborhoods with CLTs was lower than that of nongen-
trifying neighborhoods with CLTs in 2000. The results suggest that CLT units were introduced into
neighborhoods with lower housing values than those of surrounding areas, although housing prices
in gentrifying neighborhoods significantly increased throughout the last decade.
In nongentrifying neighborhoods, housing price index increased in neighborhoods with CLTs,
whereas no such change occurred in neighborhoods without a CLT. Therefore, CLTs may contribute
to increases in housing price within nongentrifying neighborhoods.
In summary, though gentrification appears to facilitate an increase in housing price, CLTs do not
produce significant effects on housing price during the gentrification process. However, CLTs
stabilize the excessive increase in housing prices in gentrifying neighborhoods by initially entering
areas with lower housing prices. In addition, among nongentrifying neighborhoods, CLTs help
increase housing prices, though not to the degree that they do in gentrifying neighborhoods.
Discussion: CLTs can moderate the negative effects of gentrification
Although a few researchers and practitioners have studied CLTs based on literature and experience
(Davis, 2010; Gray, 2008; Paterson & Dunn, 2009), their studies did not focus on the impact of CLTs
on neighborhoods. Moreover, few quantitative studies on CLTs have been published. This study
therefore seeks to address the missing quantitative evaluation of CLTs and their effects on neighbor-
hoods, with special attention paid to the process of gentrification.
JOURNAL OF URBAN AFFAIRS 13
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

This study explored whether CLTs produce an impact on gentrification and how such a
mechanism might work. To respond to the “whether”question, a binomial logistic regression
analysis was employed, the results of which showed a clear inverse relationship between CLTs and
gentrification, suggesting that CLTs may slow or dampen the negative impacts of gentrification.
We used paired t-tests to examine differences among nine relevant indicators over time between
gentrifying and nongentrifying neighborhoods, with and without CLTs. The findings were synthetically
interpreted by each indicator based on whether they mitigated the negative effects of gentrification. The
direction of CLT effects on their neighborhoods was illustrated as positive (+) or negative (−), even
though CLTs produced positive effects in most indicators. The magnitude of the effects was labeled
“consistent”when the research hypotheses were consistently supported by the paired t-test. When there
was either no support or mixed findings based on the paired t-test, the magnitude was labeled “mixed,”
because no evidence was reported on which to reject the CLT effects. Each interpretation was compiled
into Table 10 to summarize the effects of CLTs on gentrification.
Table 10 demonstrates that the effects of CLTs in slowing gentrification were significant in most
indicators and that such effects appeared more prominently in gentrifying neighborhoods than in
average neighborhoods. Taken together, we found strong support for the ability of CLTs to help
maintain middle-class ratios, education levels, and owner-occupied housing rates, as well as increas-
ing affordability in neighborhoods that were gentrifying. Moreover, CLTs increased racial diversity
and stabilized income levels and housing prices. However, contrary to the predictions of the research
hypothesis, CLTs yielded negative effects on length of residence in gentrifying neighborhoods, likely
caused by the additional residential options provided by CLT units. In addition, negative effects on
the stabilization of housing price in average neighborhoods might have been positive; considering
CLTs’positive effects on restricting excessive increases in housing price within gentrifying neighbor-
hoods, a moderate increase in housing price in an average neighborhood would have corresponded
to an increase in residents’assets.
Analysis of the results according to the negative effects of gentrification shows strong support for
the effectiveness of CLTs in counteracting lower affordability in gentrifying neighborhoods. Such
counteraction of displacement in gentrifying neighborhoods was supported by most indicators, and
CLTs produced positive effects on stabilizing housing prices in gentrifying neighborhoods.
However, CLTs’general effects on their neighborhoods were significant except for three indica-
tors: increasing the length of residence, decreasing age fluctuation, and stabilizing housing prices.
Furthermore, the high impact of CLTs on increasing affordability relates closely to their main
purpose—thus, this result supports the practical effectiveness of the CLT model.
Implications
This research provides a comprehensive perspective on the impact of CLTs on their surrounding
neighborhoods. Our population included the full set of CLTs in service in the United States, and our
response rate suggested that our results were representative of the population. A research design with
Table 10. Direction and magnitude of CLTs’effects on neighborhoods.
Negative effects of gentrification CLTs’counteracting effects on gentrification
Gentrifying
neighborhood
Average
neighborhood
Displacement Increase racial diversity +, Mixed +, Consistent
Maintain middle-class ratio +, Consistent +, Mixed
Stabilize income level +, Mixed +, Consistent
Maintain education level +, Consistent +, Mixed
Increase length of residence −(Negative) None
Decrease age fluctuation None None
Lower affordability Increase affordability +, Consistent +, Consistent
Decrease owner-occupied housing rate +, Consistent +, Mixed
Skyrocket of property price Stabilize housing price +, Mixed −(Negative)
14 M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

a pretest and posttest, as well as treatment and control groups, indicated the robustness of our
results, allowing for causal attribution to CLTs as catalysts for change (or stability) within their
neighborhoods. Previous research has revealed a range of disadvantages associated with gentrifica-
tion, but until now the role of CLTs in these circumstances has only been assumed or suggested.
Furthermore, past studies have not evaluated the actual effects of CLTs on the gentrifying process.
Therefore, this research contributes to planning theory by providing practical evidence to bolster the
assumptions of previous research.
The findings highlight the importance of using CLTs as a means to alleviate the negative impacts
of gentrification in neighborhoods. Previous literature has suggested the production and preserva-
tion of affordable housing, the building of community assets, and the intervention of government
(Henig, 1980; Kennedy & Leonard, 2001; Levy et al., 2007) as primary tools for addressing the
gentrification of areas. Our results indicate that CLTs may be one of the best ways to stabilize
neighborhoods, preserve affordability, and build community assets in neighborhoods. It warrants
much wider use than is currently seen across the United States.
Another lesson to draw from this research relates to the subcategories of CLT benefits. That is,
CLTs build community assets in neighborhoods by strengthening their stability. Neighborhood
stability, in turn, yields benefits to residents and to local jurisdictions themselves, whether they are
gentrifying or not (Rohe & Stewart, 1996). Any local government that intends to improve its
community may consider using the CLT model. CLTs are effective in increasing racial diversity
and affordability, as well as stabilizing the average household income in their neighborhoods.
Additionally, CLTs maintain middle-class ratios, education levels, and owner-occupied housing
rates. Thus, policymakers and/or community representatives should strongly consider using the
CLT model as a development tool in their neighborhoods.
Notes
1. According to Lees (2000), the terms supply-side and demand-side, economic-side and cultural-side, and/or
production-side and consumption-side are synonymous.
2. We are confident that there were no unknown CLT units in our neighborhoods, because even the locations of
CLTs that did not participate in the survey were verified.
3. Based on the definition of gentrification in this study, these five characteristics relate to demographic change (1,
2, 3, 5) and capital reinvestment (3, 4, 5).
4. The corresponding citywide area includes the original neighborhood and for this reason likely shares similar
socioeconomic characteristics. Two selection criteria were employed to identify these corresponding citywide
areas. First, they must encompass the original neighborhood. Second, the specific census places in the census
data set must have been used as corresponding citywide areas when the neighborhoods with CLTs were located
inside. If this was not the case, county data were used instead.
5. One hundred twenty percent of the AMI is commonly considered to be a threshold that determines a high-
income group. The U.S. Department of Housing and Urban Development has suggested that moderate income
groups have 80–120% of AMI in their income guidelines (Galster & Booza, 2007).
6. Migration of residents may change age distribution beyond what is ordinarily caused by demographic change
(aging, births, and deaths).
7. The affordability index is converted to help explain neighborhood affordability. A higher affordability index
indicates higher affordability.
About the authors
Myungshik Choi is an Associate Research Fellow of Housing & Land research Division at the Korea Research Institute
for Human Settlements. He holds a PhD in Urban and Regional Science Program from Texas A&M University and an
MA and BA in Urban Administration from the University of Seoul, South Korea. His areas of research interests are
land and housing policy, gentrification, community development, commons, and shared equity homeownership. His
recent research seeks to mitigate the negative effects of gentrification via the platform based on shared ownership of
local land property and building community wealth and/or assets.
JOURNAL OF URBAN AFFAIRS 15
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Shannon Van Zandt is Professor and Interim Head of the Department of Landscape Architecture & Urban Planning
at Texas A&M University. She holds a PhD in City & Regional Planning from the University of North Carolina. She
also currently holds the Nicole & Kevin Youngblood Professorship in Residential Land Development in recognition of
her scholarship on housing, real estate, and urban development. Dr. Van Zandt’s research has created a niche within
the housing and disaster fields that focuses on how the spatial distribution of residential land affects exposure, impact,
and consequences from natural disasters, particularly for socially vulnerable populations. She serves on the board of
the Texas Low-Income Housing Information Service, as well as the advisory committee of Texas Sea Grant, and has
testified before the Texas State Legislature three times over the past 2 years on issues related to housing recovery after
disaster.
David Matarrita-Cascante is an Associate Professor in the Department of Recreation, Park, and Tourism Sciences at
Texas A&M University. His research on rapid community change is guided by the field of community sociology. His
work seeks to better understand, from a sociological perspective, local processes that minimize the negative effects of
rapid change while enhancing sustainable livelihoods. A second area of interest includes the topic of human
dimensions of natural resources, guided by the fields of natural resource and environmental sociology. Along this
line, his work seeks to understand the role the natural world plays in defining human behavior as well as people’s
values and actions associated with its management and protection. His work, conducted through quantitative,
qualitative, and mixed research methods in domestic and international settings, sits at the intersection of the above
interests and various literatures including community development, amenity migration, tourism, protected areas,
natural events, and community health.
References
Abu-Lughod, J. (Ed.). (1994). From urban village to east village: The battle for New York’s lower east side. Oxford,
England: Blackwell.
Angotti, T. (2012). The gentrification dilemma. Architecture,101(8), 101–103.
Atkinson, R. (2000). Measuring gentrification and displacement in greater London. Urban Studies,37, 149–165.
doi:10.1080/0042098002339
Austin, P. C. (2011). An introduction to propensity score methods for reducing the effects of confounding in
observational studies. Multivariate Behavioral Research,46, 399–424. doi:10.1080/00273171.2011.568786
Bailey, N., & Robertson, D. (1997). Housing renewal, urban policy and gentrification. Urban Studies,34, 561–578.
doi:10.1080/0042098975925
Beauregard, R. A. (1986). The chaos and complexity of gentrification. In N. Smith & P. Williams (Eds.), Gentrification
of the city (pp. 35–55). Oxon, England: Routledge.
Betancur, J. (2002). The politics of gentrification: The case of West Town in Chicago. Urban Affairs Review,37, 780–
814. doi:10.1177/107874037006002
Betancur, J. (2011). Gentrification and community fabric in Chicago. Urban Studies,48, 383–406. doi:10.1177/
0042098009360680
Bourne, L. S. (1993). The demise of gentrification—A commentary and prospective view. Urban Geography,14,95–
107. doi:10.2747/0272-3638.14.1.95
Browne, L. E., & Tootell, G. M. (1995). Mortgage lending in Boston—A response to the critics. New England Economic
Review, (September–October), 53–72.
Brown-Saracino, J. (2004). Social preservationists and the quest for authentic community. City & Community,3, 135–
156. doi:10.1111/cico.2004.3.issue-2
Chaskin, R. J. (1997). Perspectives on neighborhood and community: A review of the literature. The Social Service
Review,71, 521–547. doi:10.1086/604277
Clay, P. L. (1979). Neighborhood renewal: Middle-class resettlement and incumbent upgrading in American
Neighborhoods.Lexington, MA: Lexington Books.
Curtin, J. F., & Bocarsly, L. (2008). CLTs: A growing trend in affordable home ownership. Journal of Affordable
Housing & Community Development Law,17, 367–394.
Davidson, M. (2011). Critical commentary. Gentrification in crisis: Towards consensus or disagreement? Urban
Studies,48, 1987–1996. doi:10.1177/0042098011411953
Davis, J. E. (1991). Contested ground: Collective action and the urban neighborhood. Ithaca, NY: Cornell University
Press.
Davis, J. E. (2006). Shared equity homeownership. Montclair, NJ: National Housing Institute.
Davis, J. E. (2007). Starting a community land trust: Organizational and operational choices (Rev. ed.). Burlington, VT:
Burlington Associates in Community Development.
Davis, J. E. (Ed.). (2010). The community land trust reader. Cambridge, MA: Lincoln Institute of Land Policy.
Defilippis, J. (2004). Unmaking Goliath: Community control in the face of global capital. New York, NY: Routledge.
16 M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Freeman, L. (2005). Displacement or succession? Residential mobility in gentrifying neighborhoods. Urban Affairs
Review,40, 463–491. doi:10.1177/1078087404273341
Freeman, L., & Braconi, F. (2004). Gentrification and displacement—New York City in the 1990s. Journal of the
American Planning Association,70,39–52. doi:10.1080/01944360408976337
Galster, G. C., & Booza, J. (2007). The rise of the bipolar neighborhood. Journal of the American Planning Association,
73, 421–435. doi:10.1080/01944360708978523
Galster, G. C., & Peacock, S. (1986). Urban gentrification: Evaluating alternative indicators. Social Indicators Research,
18, 321–337. doi:10.1007/BF00286623
Gray, K. A. (2008). Community land trusts in the United States. Journal of Community Practice,16,65–78.
doi:10.1080/10705420801977999
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, 247–265.
doi:10.1111/j.1467-9906.2011.00549.x
Hammel, D. J., & Wyly, E. K. (1996). A model for identifying gentrified areas with census data. Urban Geography,17,
248–268. doi:10.2747/0272-3638.17.3.248
Hamnett, C. (1991). The blind men and the elephant: The explanation of gentrification. In Transactions of the Institute
of British Geographers,16, 173–189.
Henig, J. R. (1980). Gentrification and displacement within cities: A comparative analysis. Social Science Quarterly,61,
638–652.
Hosmer, D. W., Jr., Lemeshow, S., & Sturdivant, R. X. (2013). Applied logistic regression (3rd ed.). Hoboken, NJ: John
Wiley & Sons.
Kennedy, M., & Leonard, P. (2001). Dealing with neighborhood change: A primer on gentrification and policy choices.
Washington, DC: Brookings Institution.
Kretzmann, J., & McKnight, J. (1993). Building communities from the inside out. Chicago, IL: ACTA.
Kutner, M. H., Nachtsheim, C. J., & Neter, J. (2004). Applied linear regression models (4th ed.). New York, NY:
McGraw-Hill Irwin.
Lang, R. E., Hughes, J. W., & Danielsen, K. A. (1997). Targeting the suburban urbanites: Marketing central-city
housing. Housing Policy Debate,8, 437–470. doi:10.1080/10511482.1997.9521260
Lees, L. (2000). A reappraisal of gentrification: Towards a “geography of gentrification.”Progress in Human Geography,
24, 389–408. doi:10.1191/030913200701540483
Lees, L., Slater, T., & Wyly, E. (2008). Gentrification. New York, NY: Routledge.
Levy, D. K., Comey, J., & Padilla, S. (2007). In the face of gentrification: Case studies of local efforts to mitigate
displacement. Journal of Affordable Housing & Community Development Law,16, 238–315.
Ley, D. (1996). The new middle class and the remaking of the central city. Oxford, England: Oxford University Press.
Logan, J. R., & Molotch, H. L. (1987). Urban fortunes: The political economy of place. Berkeley: University of California
Press.
O’Brien, R. M. (2007). A caution regarding rules of thumb for variance inflation factors. Quality & Quantity,41, 673–
690. doi:10.1007/s11135-006-9018-6
Park, Y., & Rogers, G. O. (2015). Neighborhood planning theory, guidelines, and research: Can area, population, and
boundary guide conceptual framing? Journal of Planning Literature,30,18–36.
Paterson, E., & Dunn, M. (2009). Perspectives on utilising community land trusts as a vehicle for affordable housing
provision. Local Environment,14, 749–764. doi:10.1080/13549830903096486
Redfern, P. A. (2003). What makes gentrification “gentrification”?Urban Studies,40, 2351–2366. doi:10.1080/
0042098032000136101
Rohe, W. M., & Stewart, L. S. (1996). Homeownership and neighborhood stability. Housing Policy Debate,7,37–81.
doi:10.1080/10511482.1996.9521213
Rose, D. (1984). Rethinking gentrification—Beyond the uneven development of Marxist urban theory. Environment
and Planning D - Society & Space,2,47–74. doi:10.1068/d020047
Saegert, S., & Benitez, L. (2005). Limited equity housing cooperatives: Defining a niche in the low-income housing
market. Journal of Planning Literature,19, 427–439.
Sawicki, D. S., & Flynn, P. (1996). Neighborhood indicators: A review of the literature and an assessment of conceptual
and methodological issues. Journal of the American Planning Association,62, 165–183. doi:10.1080/
01944369608975683
Smith, N. (1979). Toward a theory of gentrification—A back to the city movement by capital, not people. Journal of the
American Planning Association,45, 538–548. doi:10.1080/01944367908977002
Smith, N. (1996). The urban frontier: Gentrification and revanchist city. New York, NY: Routledge Press.
Smith, N., & LeFaivre, M. (1984). A class analysis of gentrification. In J. J. Palen & B. London (Eds.), Gentrification,
displacement, and neighborhood revitalization (pp. 43–63). Albany: State University of New York Press.
Smith, N., & Williams, P. (1986). Alternatives to orthodoxy: Invitation to a debate. In N. Smith & P. Williams (Eds.),
Gentrification of the city (pp. 1–14). Oxon, England: Routledge.
JOURNAL OF URBAN AFFAIRS 17
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017

Sullivan, D. M. (2007). Reassessing gentrification measuring residents’opinions using survey data. Urban Affairs
Review,42, 583–592. doi:10.1177/1078087406295828
Sumka, H. J. (1979). Neighborhood revitalization and displacement: A review of the evidence. Journal of the American
Planning Association,45, 480–487. doi:10.1080/01944367908976994
Wyly, E. K., & Hammel, D. J. (1999). Islands of decay in seas of renewal: Housing policy and the resurgence of
gentrification. Housing Policy Debate,10, 711–771.
Zukin, S. (1987). Gentrification: Culture and capital in the urban core.Annual Review of Sociology,13, 129–147.
doi:10.1146/annurev.so.13.080187.001021
18 M. CHOI ET AL.
Downloaded by [Korea Research Institute for Human Settlements] at 19:17 27 September 2017