Forewarned: The Use of

Neighborhood Early Warning

Systems for Gentrification
and Displacement
Karen Chapple
Miriam Zuk
University of California, Berkeley

Abstract
The 1980s saw the emergence of neighborhood early warning systems that use indica-
tors to assess patterns of neighborhood change. In more recent years, new systems and
analyses are measuring the risk of gentrification and displacement. Using information
from a dozen interviews with developers and users and from a survey conducted in one
region, we show that policymakers, community residents, and other stakeholders are
actively using these early warning systems strategically, tactically, and for empowerment.
Although the extent to which the analyses have actually caused policy shifts is unknown,
the early warning systems clearly have influenced the urban debate about housing and
neighborhood change. The durability of these efforts, however, remains an outstanding
question. Cities have not yet sought to develop these tools and strategies for more equitable,
inclusive neighborhood change, yet city government is a logical home for early warning
systems, especially given new technological capabilities.

Introduction
Neighborhoods change continually because of the movement of people and capital, both private
and public. Change is often visible, as newcomers walk the streets or buildings and infrastructure
are built and demolished. At the same time, change may be hard to discern, as property transfers
and even the arrival of new tenants are not publicized. The process may take decades to unfold and
may be nonlinear; change can stall or reverse, and the neighborhood may never fully transform.

As local residents and policymakers struggle to discern the nature and extent of changes, research-
ers have devised “neighborhood early warning systems” to describe change processes and even

Cityscape: A Journal of Policy Development and Research • Volume 18, Number 3 • 2016 Cityscape 109
U.S. Department of Housing and Urban Development • Office of Policy Development and Research
Chapple and Zuk

predict future transformation. These toolkits, which take the form of either reports or online
guides, tend to focus on economic and racial/ethnic change at the neighborhood scale via demo-
graphic and property data. The idea of early warning is that, by tracking investment, disinvest-
ment, and population flows at the local level, policymakers can design cost-effective interventions
before the pace of change accelerates and patterns become entrenched (Snow, Pettit, and Turner,
2003). In the case of neighborhood decline, early warning might mean identifying crime hotspots
or abandoned properties. For neighborhoods that are revitalizing, toolkits tend to focus on areas of
housing sales, racial transition, and new amenities, among other factors.

The first generation of toolkits from the 1980s and 1990s has now disappeared,1 but both the
overheating of the housing market and the planning of new transit systems have led to new interest
in understanding neighborhood change, specifically in the form of gentrification and displacement.
New early warning systems with an online presence have emerged in Portland, Oregon; the San
Francisco Bay Area in California; Chicago, Illinois; and Minneapolis-St. Paul, Minnesota. Many other
regions also have conducted analyses. This new generation of toolkits has the potential to transform
policies to stabilize and/or revitalize neighborhoods, especially if, this time around, they find more
permanent homes. One pathway might be to expand the “smart cities” movement beyond its cur-
rent focus on efficiency to proactive policymaking around inclusion (Pettit and Greene, 2016).

Little is understood, however, about precisely how stakeholders are using the systems and what
impact those systems have on policy. Early warning systems have complex and multiple goals in
contrast with smart cities systems, which primarily attempt to make city systems more responsive
to constituents. To make the case for integrating early warning systems into city operations, it is
important to understand their value. This article describes the intent and use of these toolkits,
assessing their ability to make policy more effective, their potential sustainability, and, for a few,
their predictive capability.

The following section discusses the evolution of urban data capabilities and then describes the first
generation of early warning toolkits. The next section presents a survey of the landscape of current
toolkits, including the Urban Displacement Project tool in the San Francisco Bay Area, which the
authors developed. The next section, using information from a dozen interviews with developers
and users and also from a survey conducted in one region, explores the different ways that toolkits
have been used. The final section lays out next steps for system development, suggesting ways to
increase the relevance of toolkits to the planning and development decisions that elected officials
and communities face.

1
Snow, Pettit, and Turner (2003) profiled four early warning systems: the Chicago Neighborhood Early Warning System,
by the Center for Neighborhood Technology; Neighborhood Knowledge Los Angeles, at the University of California, Los
Angeles Center for Neighborhood Knowledge; the Philadelphia Neighborhood Information System, at the University of
Pennsylvania; and the Minneapolis Neighborhood Information System, at the University of Minnesota. Each has either
disappeared or not been updated in many years.

110 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

Perspectives on Smart Cities, Neighborhood Change, and

Early Warning Toolkits
The current generation of neighborhood early warning systems dates from the emergence of Geo-
graphic Information Systems (GISs). A movement to democratize data resulted in broad experi-
mentation with data portals that characterize neighborhood change. Most recently, the movement
has shifted focus to making cities smarter.

The Use of Data and Maps in Cities and Neighborhoods

Shortly after GISs became widely available on personal computers in the early 1990s, a set of
intermediaries emerged to create more democratic access to data; many of these intermediaries
were part of the Urban Institute’s National Neighborhood Indicators Partnership (NNIP; Treuhaft,
2006).2 These intermediaries, often community-based organizations working in partnership with
universities, gather neighborhood-level data, organize them into a database, and help community
actors map and analyze the data by themselves. The focus thus was on empowerment, building trust
and capacity in communities that historically had been on the wrong side of the map (for example,
through practices such as redlining) (Treuhaft, 2006).

The movement to democratize data has recently morphed into interest in smart cities, which op-
timize urban systems and service delivery through real-time monitoring and control. The promise
of smart cities is that new digital tools that aid in the collection, analysis, and dissemination of data
will help cities shift from a compliance mode to a problem-solving mode (Goldsmith and Craw-
ford, 2014). At the same time, it is believed, technology will strengthen civil society as constituents
coproduce solutions with government (Goldsmith and Crawford, 2014). Absent, however, from
smart cities experiments is the application of technology to more equitable outcomes, particularly
in neighborhoods, and also the input from community organizations (Baud et al., 2014; Pettit and
Greene, 2016).

Despite the enthusiasm about moving toward smarter cities and more democratic data, questions
remain about how the data and maps produced are actually used. Data analysis and maps either
remain for internal use in decisionmaking, whether by government agency or community organi-
zation, or they are made available to external audiences to garner attention or generate new ideas.
Users, particularly community groups, may use maps in a strategic way (for example, to identify
needs or target resources), as a tactic to raise awareness or implement solutions, for administration
(for example, for service delivery), for organizing or building the capacity of a constituency, or sim-
ply for exploration to see if spatial knowledge legitimizes local experience or raises questions about
city policy (Craig and Elwood, 1998; Ghose, 2011). Over the long term, GIS analysis and maps
are thought to have the potential to transform planning, policy, and programs (Ramasubramanian,
2011)—yet, little systematic evidence supports this thinking.

2
NNIP, founded in 1996, consists of a loose network of data intermediaries in 30 cities.

Cityscape 111
Chapple and Zuk

The Rise of Neighborhood Early Warning Systems

Scientists and social scientists alike have long coveted the ability to predict the future. As the avail-
ability of new data has made it possible to identify the factors predicting or simply has correlated
with different phenomena, researchers have tried to use these indicators to predict future change.
Thus, early warning systems are now available for crime hotspots and gang homicides (Gorr and
Lee, 2015; Sampson, 2011), housing abandonment and foreclosure (Hillier et al., 2003; Williams,
Galster, and Verma, 2013), housing price appreciation (Galster and Tatian, 2009), land use change
(Waddell, 2002), and even tornados (Oleske, 2009).

The first neighborhood-level early warning system was pioneered beginning in 1984 by the Center
for Neighborhood Technology in Chicago. The idea was to create a portal of property data, such as
information on tax delinquencies, code violations, and utility shutoffs, which could then be used
to monitor neighborhood housing conditions (and thus spur intervention). Because many forms of
financial disinvestment are invisible, identifying patterns in a timely manner can be preventative.
An early Urban Institute report describing four such systems (in Los Angeles, California; Minneapolis;
and Philadelphia, Pennsylvania—in addition to Chicago) found that they all provided indicators of
financial disinvestment based on parcel-based data—aggregated in different ways, depending on
the issue—obtained from the local government (Snow, Pettit, and Turner, 2003). With an audience
of government agencies and community-based organizations, the systems were disseminated on
the web and housed at academic or research institutions.

Cities and other stakeholders are interested in monitoring neighborhood decline for immediate
reasons—the potential that families will lose their shelter—and for long-term issues—particularly
the spiral of decline that can result in a variety of costly impacts for families and cities alike (Wil-
son, 1987). By contrast, the rationale for monitoring neighborhood revitalization or gentrification
is murkier.

Gentrification is a simultaneously spatial and social practice that results in “the transformation of
a working-class or vacant area of the central city into middle-class residential or commercial use”
(Lees, Slater, and Wyly, 2008: xv)—meaning the influx of both capital (real estate investment) and
higher-income or higher-educated residents. Displacement—when households are forced to move
out of their neighborhood—can be a negative outcome of gentrification but may also precede it
(Marcuse, 1986). Real estate investors, including prospective homebuyers, certainly take an inter-
est in gentrification. For cities, it is important to understand neighborhood upgrading not only to
stabilize communities but also to intervene proactively before intervention (for example, mitigating
displacement) becomes costly and difficult (Pettit and Greene, 2016).

In one earlier iteration of work predicting gentrification—a presentation by researchers from the
Urban Institute (Turner and Snow, 2001)—the researchers characterized the process of gentrifica-
tion by (1) shift in tenure, (2) increase in downpayment and decrease in FHA financing, (3) influx
of households interested in urban living, and (4) increase in high-income-serving amenities such
as coffee shops or galleries. Analyzing data for the DC area, they identified the following five
predictors of future gentrification (defined as sales prices that are above the DC average) in low-
priced areas: (1) adjacency to higher-priced areas, (2) good access to the Metro subway system,

112 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

(3) historic architecture, (4) large housing units, and (5) more than 50 percent appreciation in
sales prices between 1994 and 2000. Census tracts were scored for each indicator and then ranked
according to the sum of indicators, with a maximum value of 5.

In 2009, the Association of Bay Area Governments sponsored an analysis of neighborhood change
in the San Francisco Bay Area from 1990 to 2000, which predicted neighborhood susceptibility to
gentrification, with a disclaimer that it was not possible to measure resident displacement via this
method (Chapple, 2009). Chapple adopted Freeman’s (2005) definition of gentrifying neighbor-
hoods as low-income census tracts in central city locations in 1990 that, by 2000, had experienced
housing appreciation and increased educational attainment that were higher than the nine-county
regional average and then constructed a multivariate statistical model that had gentrification as
the dependent variable and a set of 19 socioeconomic, locational, and built environment factors
for 1990 as independent variables. When census tracts scored above the regional average for each
variable, they received a value of 1; the susceptibility index summed the scores across the variables.

In 2011, Atkinson et al. characterized household vulnerability to displacement from neighbor-

hoods that gentrified between 2001 and 2006 in the Melbourne and Sydney, Australia, greater
metropolitan areas. A vulnerability score (from 1 to 13) was measured based on tenure, number of
employed people per household, and occupation. Displacement rates were calculated by dividing
the number of out-migrants with vulnerability characteristics by the number of households with
these characteristics exposed to the likelihood of moving in 2001. Neighborhoods that had higher-
than-projected numbers of high-income, owner-occupant, and professional populations were
designated gentrified.

Researchers have used myriad indicators and sources of data for characterizing residential gentrifi-
cation displacement, each with its own set of advantages and disadvantages (exhibit 1). The table
in exhibit 1 summarizes quantitative data sources only; however, data on many of the drivers and
impacts of gentrification and displacement are not regularly gathered or are difficult to quantify.

Exhibit 1
Indicators and Data Sources for Analyzing Gentrification and Displacement (1 of 2)
Indicator Type Indicators Data Sources
Change in property Sales value, property value County tax assessors’ offices, finance
values and rents departments, data aggregators
Rent Data aggregators, apartment operating
licenses, craigslist
Changes in availability of restricted HUD, housing departments
affordable housing

Cityscape 113
Chapple and Zuk

Exhibit 1
Indicators and Data Sources for Analyzing Gentrification and Displacement (2 of 2)
Indicator Type Indicators Data Sources
Investment in the Building permits, housing starts, Jurisdictions’ building or planning
neighborhood renovation permits, absentee departments
ownership
Mortgage lending and HMDA and assessors’ data
characteristics
Sales (volume and price) County assessors’ offices, data
aggregators
Condominium conversions Assessors’ offices, housing departments,
public works departments
Change in community and business Chambers of Commerce, Dun &
organizations (for example, Bradstreet, neighborhood or local
number, membership, nature of business associations, and so on
activities)
Public investments (for example, Public works departments, transit
transit, streets, parks) agencies, parks and recreation
departments, and so on.
Disinvestment Building conditions, tenant Surveys, censuses, maps, building
complaints, vacancies, fires, departments, utility shut-off data, fire
building condemnation departments
School quality, crime, employment Departments of education, police
rates, neighborhood opportunity departments/crime maps, censuses,
Bureau of Labor Statistics
Neighborhood quality Local surveys
Change in tenure Tenure type, change in tenancy Building departments, assessors’
and demographic offices, censuses
changes Evictions Rent boards, superior courts
Foreclosure HUD, proprietary data sources
Demographics data on in- vs. out-­ Censuses, voter registration data, real
movers (for example, race, ethnicity, estate directories, surveys, American
age, income, employ­ment, educa­ Housing Survey, departments of
tional achievement, marital status) motor vehicles
Investment potential Neighborhood and building Tax assessors, censuses, deeds, and
characteristics (for example, age so on
and square footage, improvement-
to-land ratio)
Neighborhood perceptions Surveys of residents, realtors,
lenders, neighborhood businesses,
newspapers, television, blogs, and
so on
Reasons that people Reason for move Surveys of in-movers and out-movers,
move in or out of state housing discrimination
neighborhood complaints database
Coping strategies and Crowding or doubling up Censuses, utility bills, building
displacement impacts footprints
Increased travel distance and time Censuses
HMDA = Home Mortgage Disclosure Act. HUD = U.S. Department of Housing and Urban Development.

114 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

The Future of Neighborhood Early Warning Systems

More than 30 years after the first neighborhood early warning system emerged, those systems arguably
have failed to meet their potential. In fact, the first early warning systems for neighborhood decline
have not survived the test of time. Although more research would be necessary to determine why,
three explanations seem likely: (1) all the systems were housed at nonprofit organizations or uni-
versities, where changes in personnel and leadership can change institutional focus (as opposed,
for example, to a city, which has a more constant mission); (2) all the systems relied primarily on
funding from philanthropy, which changes its focus frequently, and/or the U.S. Department of
Housing and Urban Development (HUD), which has experienced repeated budget cuts in the past
few decades; and (3) none of the systems developed a broad base of users (beyond community-
based organizations).

The first generation of early warning systems innovated new uses of local data and offered consid-
erable promise to shape policymaking (Snow, Pettit, and Turner, 2003). The lack of sustainability in
these systems, however, suggests that they failed to convince potential users about the importance
of early warning and preventive approaches to neighborhood change. Moreover, three decades
after the first research on gentrification and displacement, we continue to struggle to predict which
neighborhoods will gentrify and who will benefit (and suffer). Most of the debate about gentrifica-
tion and displacement has remained in academic spheres, outside of the policy realm—until the
recent arrival of warning systems for gentrification and displacement.

The emergence of the smart cities movement suggests the potential of these tools. Research suggests
that data on gentrification and displacement underrepresents the most disadvantaged populations
and presents a mismatch between data and lived experience (Zuk et al., 2015). This underrepre-
sentation might be overcome by user-generated geographic content, volunteered by residents and
posted via interfaces like Flickr (Goodchild, 2007). With better data, prediction might improve,
and, with more accessible portals, different stakeholders may coproduce more effective policies.
Pettit and Greene (2016) envision the following—

But what if city leaders and community groups could get ahead of these changes and act
early to direct neighborhood changes toward more inclusive outcomes? Using big data
and predictive analytics, they could develop early warning systems that track key indica-
tors of neighborhood change and predict future trajectories (Pettit and Greene, 2016: 2).

The next section presents an overview of how the next generation of early warning systems is faring.

Neighborhood Early Warning Systems: Surveying the

Landscape
To examine further the use of early warning systems for neighborhood change—and gentrification
and displacement in particular—we next establish the universe of systems via a web scan. Two
starting points were the Urban Institute’s NNIP and the Obama administration’s open data portal,
The Opportunity Project. We also searched the web on terms such as “neighborhood,” “gentrifica-
tion,” and “displacement” and asked our interviewees for systems we had missed.

Cityscape 115
Chapple and Zuk

We identified three types of websites that explore neighborhood issues: (1) neighborhood indica-
tor maps (typically of development, such as local educational attainment or housing construction,
or quality of life, often represented by amenities), (2) opportunity maps, and (3) racial/economic
change maps (including gentrification).3 To narrow our focus, we chose just the sites focusing
on gentrification within this last category, which included projects in Chicago, Minneapolis-St.
Paul (two projects), Portland, San Francisco, and Washington, DC. We excluded several sites that
depict neighborhood change without an explicit focus on gentrification or assessment of risk.4 We
then added projects from several cities—Charlotte, North Carolina; Houston, Texas; Los Angeles;
Seattle/Puget Sound, Washington; and St. Louis, Missouri—that had produced recent assessments
of gentrification or displacement risk with a report, rather than a web interface, as the final prod-
uct. Again, we excluded recent gentrification reports that were not framed as risk assessments.5

From the 11 projects, we interviewed 9 of the system creators and attended a presentation of 1;
the last site is our own. Most of the interviews occurred via telephone and lasted 45 to 60 minutes,
using a semistructured format; one interview was by e-mail. The analysis also draws from a survey
of users (n = 33) of the University of California, Berkeley’s Urban Displacement Project toolkit.6

The projects generally fall into two broad categories: (1) those developed by universities, with
online map interfaces, and (2) those developed by cities as reports for internal use (exhibit 2). Per-
haps because of the role of city government in many of the projects, most of the analyses examine
neighborhood change within city, rather than regional, limits. The most common audience, both
intended and actual, is city government and community organizations; others specified regional
agencies, community members, and elected officials as their target audience. All the sites rely
primarily on U.S. census data at the tract level, typically using the data with standardized census
tract boundaries provided by GeoLytics, Inc., or Brown University. Most of the projects span at
least two decades (1990 to 2010 or 1990 to 2014), and two projects (Chicago and St. Louis) use
1970 as the starting year. Two sites (Portland and San Francisco) also add parcel-based data on
recent home sales, and two (San Francisco and Washington, DC) add data from the U.S. Census
Bureau’s Longitudinal Employer-Household Dynamics program on job accessibility and also add
a rail transit station layer. One site (San Francisco) also uses data about amenities (parks, transit,
walkability), property characteristics (from the tax assessor), and nonprofit organizations.

3
Our scan identified 24 of these websites, but we suspect that many more exist.
4
These sites include HUD’S Affirmatively Furthering Fair Housing Assessment Tools (huduser.gov/portal/affht_
pt.html#affhassess-tab); Code for Boston’s Ungentry (http://codeforboston.github.io/ungentry/); and sociologist Michael
Bader’s racial/ethnic change maps for New York, Los Angeles, Chicago, and Houston (http://mikebader.net/media/
neighborhoodtrajectories/map.html?city=newyork).
5
These reports include the 2016 New York University Furman Center for Real Estate and Urban Policy annual report on
New York City housing (http://furmancenter.org/research/sonychan) and two reports on Philadelphia by the Federal Reserve
Bank (Ding, Hwang, and Divringi, 2015) and the Pew Charitable Trusts (2016).
6
The Urban Displacement Project solicited survey responses from a list of 395 stakeholders in the nine-county Bay Area,
including housing policy advocates, planning directors, and elected officials. After two e-mail solicitations, the project
received 33 responses (a response rate of 8 percent). The survey asked users 10 questions about how they used the site
(maps, case studies, and policy inventory) and also asked how the site could be improved.

116 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

Exhibit 2
Neighborhood Early Warning Systems for Gentrification and Displacement (1 of 2)
Policy
Type of Geog-
City/Region Host Goal Users Format Influ-
Project raphy
ence?
Charlotte, Neighbor- City of City Tactical: Understand City, some Internal NA
North hood change Charlotte how to do equitable community report
Carolina analysis and inclusive devel- organiza-
opment tions
Chicago, Gentrification University City Tactical and em- Community Report Yes
Illinois index of Illinois powering: Measure organiza- and maps
at Chicago change and provide tions on line
tools
Houston, Gentrification Local City Strategic, tactical, LISC, Internal Yes
Texas index and at- Initiatives empowering: Use community report
risk indicator Support as advocacy tool for organiza-
Corpora- LISC tions
tion
Los Angeles, Gentrification City of Los City Strategic: Help city Mayor's Internal NA
California index Angeles target initiatives office report
within a large grant
program
Minneapolis, Housing University Twin Strategic, tactical, Community Report Yes
Minnesota ­market index of Minne- Cities empowering: Start a organiza- and maps
sota Twin conversation, inform tions, city on line
Cities policymakers and
residents
­Minneapolis-­ Gentrification Minnesota Region Strategic, tactical: Community Report No
St. Paul index and at- Center for Spark conversation, organiza- and inter­
­metropolitan risk indicator Environ- implement mitiga- tions active
area, mental tions, obtain funding maps on
­Minnesota Advocacy line
Portland, Gentrification Portland City Tactical, empower- City, com- Report Yes
Oregon index and at- State ing: Show where munity and maps
risk indicator University gentrification is hap- organiza- on line
(hosted by pening in Portland tions
The Or-
egonian)
St. Louis, Index of University City Tactical and empow- Community Report on Yes
Missouri "neighbor- of ering: Show which organiza- line
hood vitality" Missouri– neigborhoods are tions
St. Louis "rebounding"
San Francisco Gentrification ­University Region Tactical and em- Local gov- Report Yes
Bay Area, index and at- of powering: Describe ernment, and inter­
California risk indicator California, current patterns community active
Berkeley of neighborhood organiza- maps on
change and city tions, line
policies elected
officials

Cityscape 117
Chapple and Zuk

Exhibit 2
Neighborhood Early Warning Systems for Gentrification and Displacement (2 of 2)
Policy
Type of Geog-
City/Region Host Goal Users Format Influ-
Project raphy
ence?
Seattle- Neighborhood Puget Region Strategic and Local gov- Report on Yes
Puget Sound, typology Sound tactical: Provide ernment, line
Washington and at-risk Regional jurisdictions a tool for community
indicator Council station-area plans members
Washington, Gentrification Univer- Region Tactical: Understand Local gov- Maps on NA
DC index and at- sity of change primarily ernment line
risk ­indicator Maryland, around transit (Purple
(not yet College Line on the Metrorail
released) Park system)
LISC = Local Initiatives Support Corporation. NA = not applicable.

Analyzing Risk
The first generation of reports analyzing gentrification and displacement risk generally all followed
the same methodology; that is, run correlations or regressions to identify predictors of gentrifica-
tion and/or displacement and then assign each factor a value to come up with a susceptibility score
(Atkinson et al., 2011; Chapple, 2009; Turner and Snow, 2001). The analyses behind the current
set of early warning systems—in Chicago, Houston, Portland, the San Francisco Bay Area, and
Seattle/Puget Sound, as described further below—have improved on this methodology by looking
at the dimension of time (that is, past and present neighborhood change dynamics in addition to
the extent of vulnerability). Many analyses also make a useful analytic distinction between gentrifi-
cation and displacement, while still analyzing both.

The Chicago gentrification index (Nathalie P. Voorhees Center, 2014) determined relevant factors
based on a literature review. It provided a “score” for each “community area” in 1970, 1980, 1990,
2000, and 2010, based on a composite index that compares the community area to the city at large
for 13 indicators. Then, a neighborhood change typology (displayed in maps) was constructed
not just from these scores but also from their change between 1970 and 2010. A separate toolkit
identified housing, land use, and other tools appropriate for each of three stages: (1) before gentri-
fication, (2) midstage gentrification, and (3) late-stage gentrification (Nathalie P. Voorhees Center,
2015).

Building off the same methodology as Chapple (2009), Local Initiatives Support Corporation
(LISC) researchers constructed a model predicting gentrification in neighborhoods of Houston,
using a slightly narrower definition of gentrifying neighborhoods (Winston and Walker, n.d.). The
LISC researchers used the regression coefficients and continuous independent variables in predict-
ing susceptibility to gentrification.

In Portland, Bates (2013) predicted market changes based on vulnerability to displacement, demo-
graphic changes, and housing market conditions, a method that was replicated in the Twin Cities
gentrification risk assessment performed by the Minnesota Center for Environmental Advocacy.
Tracts were vulnerable to displacement in 2010 when they had higher-than-average populations

118 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

of renters and communities of color, few college degrees, and lower incomes. For housing market
conditions, Bates defined neighborhood market typologies as (1) adjacent tracts (low/moderate
2010 value, low/moderate appreciation, next to high-value/appreciation tract); (2) accelerating
tracts (low/moderate in 2010 with high-appreciation rates); and (3) appreciated tracts (low/moderate
1990 value, high 2010 value, high 1990 to 2010 appreciation). Combining this information with
demographic shifts for vulnerability factors between 2000 and 2010, the study identified six
neighborhood types ranging from early to mid- to late-stage gentrification. Bates then used these
typologies to recommend how to tailor policy approaches to the specific characteristics and needs
of neighborhoods.

In the San Francisco Bay Area, the Urban Displacement Project provided a typology analysis that
characterizes Bay Area neighborhoods (census tracts) according to their experience of gentrifica-
tion and risk of displacement.7 This early warning system was based on a gentrification index
that adapts the methodologies of various researchers (for example, Bates, 2013; Freeman, 2005;
Maciag, 2015) to characterize places that historically housed vulnerable populations but have since
experienced significant demographic shifts and real estate investment.

The loss of low-income households between 2000 and 2013 was used as a proxy for displacement.
On average, Bay Area census tracts’ low-income population grew by 59 households between 2000
and 2013. The typology therefore assumes that any neighborhood that experienced a net loss of
low-income households while stable in overall population is a result of displacement pressures.8
After constructing regression models to estimate the predictors of both gentrification and loss of
low-income households/displacement, the project developed place typologies for risk of either
gentrification-related displacement or exclusion-related displacement (which occurs in higher-
income neighborhoods). Unlike the other studies, results were vetted via several workshops with
a project advisory committee and also via community forums. Based on these interactions, tracts
were divided into low-income and moderate- to high-income tracts to capture the displacement
pressures occurring in nongentrifying neighborhoods that are also losing low-income households.
Exhibit 3 presents the resulting typology. The Urban Displacement Project’s website also includes
an inventory of policies available in each jurisdiction (exhibit 4).

The Puget Sound Regional Council project, conducted with the Center for Transit-Oriented
Development, used descriptive methods to construct a typology of neighborhoods based on risk
factors (the “people profile”) and market strength (the “place profile”), which then formed the basis
for suggesting policy responses (PSRC, 2013). For the people profile, one axis consisted of social

7
This project was a side product of a larger study funded by the California Air Resources Board and the Metropolitan
Transportation Commission (via HUD’s Sustainable Communities Initiative) that involved extensive qualitative and
quantitative regional analysis to better understand the nature of neighborhood change and displacement in the Bay Area
and their relationship to transit.
8
We assume that a tract that lost low-income households during this period underwent some process of displacement when
combined with other indicators such as a loss of market-rate affordable units or a decline of the in-migration of low-income
population into that tract beyond the regional median. Although the change in low-income households could be because
of income mobility (for example, low-income households moving into middle- or upper-income categories, or vice versa),
from our analysis of data from the Panel Study of Income Dynamics, we estimate that there would have been a net increase
in low-income households in most places likely because of the Great Recession (December 2007 to June 2009); therefore,
our estimates of displacement are likely an underestimate.

Cityscape 119
Chapple and Zuk

Exhibit 3
Displacement/Gentrification Typologiesa
Lower-Income Tracts (> 39% of households are Moderate- to High-Income Tracts (< 39% of
considered low income) households are considered low income)
Not losing low-income households or very early Not losing low-income households or very early
stages stages
• Does not fall within any of the following catego- • Does not fall within any of the following catego-
ries ries
At risk of gentrification or displacement At risk of displacement
• Strong market • Strong market
• In TOD • In TOD
• Historic housing stock • Historic housing stock
• Losing market-rate affordable units • Losing market-rate affordable units
• Employment center • Employment center
Undergoing displacement Undergoing displacement
• Already losing low-income households and • Already losing low-income households
naturally affordable units • Either naturally affordable units or in-migration of
• In-migration of low-income residents has low-income residents has declined
­declined • Stable or growing in size
• Stable or growing in size
Advanced gentrification Advanced exclusion
• Gentrified between 1990 and 2000 or between • Very low proportion of low-income households
2000 and 2013 based on— • Very low in-migration of low-income households
žžNeighborhood vulnerability
žžDemographic change
žžReal estate investment
TOD = transit-oriented development.
a
Tracts with 0 population in 2010 were excluded from the analysis (8 tracts). In addition, tracts where more than 50 percent
of the population in 2010 was in college were excluded from the analysis (11 tracts).

Exhibit 4

Policy Inventory on Urban Displacement Project Website

Source: http://www.urbandisplacement.org

120 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

infrastructure and access to opportunity. The second axis—change/displacement—measured risk

of displacement resulting from recent neighborhood change, current community risk factors, and
current and future market pressure. The place profile also consisted of two dimensions: (1) urban
form that supports a dense and walkable transit community and (2) the likelihood that the com-
munity will change in response to real estate market strength. Combining the people and place
typologies, they identified eight general typologies; for each typology, they identified implementa-
tion and policy approaches.

Thus, in an attempt to predict change more accurately, early warning systems and related projects
are gradually improving in methodology. Notable methodological shifts include the analysis of
multiple stages of both gentrification and displacement, building on the approach of Bates (2013);
the shift to a regional, rather than municipal, framework; and the mixing of quantitative and
qualitative approaches. Conceptualizing gentrification and displacement as a long-term, multistage
process, rather than a binary state or on/off switch, has helped build local buy-in into the early
warning systems. Looking at many different cities within a region helps localities understand
regional housing market dynamics and learn about different policies. Checking results with local
residents and key informants helps ensure that the maps represent conditions on the ground.

Methodological problems remain, however, particularly in terms of the predictive ability of the
models. Methods are still far from transparent: models are not readily replicable, and the scores
can be hard to understand. The next section describes how stakeholders are using the models in
practice and also the effectiveness of the new approaches.

The Use and Impact of Neighborhood Early Warning Systems

This section examines the use and impact of these projects, looking at those that assess gentrifica-
tion and/or risk (in Chicago, Houston, Minneapolis-St. Paul, Portland, San Francisco, St. Louis)
and also examines the other neighborhood change reports (in Charlotte, Los Angeles, Minneapolis,
and Seattle/Puget Sound).9 We assess first how internal actors, and then external stakeholders, use
early warning systems. We then examine what impact the projects have had on policymaking and
how accurate they are at predicting change.

Internal Use
One obvious use for early warning analyses is in strategic planning for housing and neighbor-
hoods. Maps that show how neighborhoods are changing and that anticipate future change can
help stakeholders bring attention to imminent problems and target resources. If the map suggests
that change is in very early stages, the neighborhood can strategize about actions to take during the
long term; for example, the Houston systems architect said, “…in Houston, we are a few years or a
decade behind other metropolitan areas in terms of the waves of gentrification and things coming.
So what we realized is that by doing research now, we could get ahead of that.” The gentrification
analysis showed where change was anticipated yet land was still cheap, so that intermediaries
could target land acquisition funds strategically.

9
Because the Washington, DC site has not yet been launched, it is too early to assess its use and impact.

Cityscape 121
Chapple and Zuk

The Houston project was strategic, not just in terms of timing but also in policy approach and
ownership. Before the analysis, stakeholders had expressed some disagreement about how to
spend disaster recovery money. Having the data helped advocates to say, “…if we’re doing this
investment, let’s also create and preserve affordable housing opportunities in places at risk of
gentrification”—but without making enemies by specifically endorsing certain policies within the
report itself.

The Houston system creator said—

We had a strategically placed piece of analysis that could help community stakeholders
on our side make a point about what policy ought to be. Not a distraction, not something
that came out of Washington, DC, saying this is what y’all ought to do. Because that
would have been suicidal.

Another strategic, internal use of maps is targeting resources, as with the Housing Market Index
(HMI) in Minneapolis, which helps determine the blocks where funds to fix vacant property can be
most effectively spent. One developer on the Minneapolis project said, “It has been very, very, very
useful.... When you’re involved in politics, and competition for scarce resources, the more facts
you can provide, the better you are. The HMI are facts. And that speaks much louder than any
political will.”

In Charlotte, where the use of the report remained internal to the city government, the analysis
became a tactic to broaden the framework and discussion of neighborhood change. The initial
referral from the city council had been to look at gentrification, but instead the city “looked more
broadly at neighborhood change and the challenges that can arise in the context of gentrification
across all neighborhoods, plus the close ties that this issue has with economic opportunity and the
historic patterns of economic and racial segregation in Charlotte—consequently, we looked at a
broad range of indicators.”

The analysis in Charlotte ultimately supported the development of a much broader housing strat-
egy than anticipated, with a wide array of tools and strategies to manage neighborhood change.

Once the analysis is in place, it can create its own momentum. In Seattle, the Puget Sound Regional
Council analysis established—after considerable debate with advocates—that four neighborhoods
in southeast Seattle were at high risk. Years later, planners working on the update to the Seattle
Comprehensive Plan used the analysis as a background document to show that the community was
at risk. Developers of the gentrification typology in Minneapolis-St. Paul have a similar intent—to
create the momentum to fund and implement the mitigations for neighborhood preservation and
equitable development in St. Paul’s Central Corridor Development Strategy.

External Use
The most common use of early warning indicators and maps is as a tactic to spark a conversa-
tion, generate new ideas, or show how to implement solutions. The survey of users of the Urban
Displacement Project in the San Francisco Bay Area suggested that this was the primary use of that
warning system. Users volunteered that it was a tool to start dialogue: “I’ve used the maps to show
policymakers that my neighborhood is at risk of displacement.”

122 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

In the Bay Area, the tool also serves to legitimize other work—
My organization provides legal research, advice, education, and advocacy to support
communities in developing community-owned economic structures. This data has been
useful in better understanding the dynamics of displacement internally, as well as in
communicating about the importance of our work to the public.

It also lets advocates know where cities lack antidisplacement policies, so they can push for imple­
mentation. A user of the Urban Displacement Project in the Bay Area reported using the site “to assess
which areas have been most impacted in order to identify mitigation strategies for nonprofits that
lease in those areas.” Users reported using it “to check on what policies have been implemented
by Bay Area jurisdictions to produce more housing” and “assessing opportunity for preservation
strategies and making the case for funding.” Because the maps are regional, advocates use them to
advocate at the regional level: “[We] identify which cities are performing well and which are not.
[We] advocate for MTC [Metropolitan Transportation Commission] to use this info to guide funding
through OBAG [a regional grant program to encourage density] to incentivize better local policies.”

In St. Louis, the release of the index of “neighborhood vitality” also brought new attention to
“rebounding” neighborhoods, helping to spark a conversation about how reinvestment occurs.
Researchers at the University of Missouri–St. Louis sponsor a morning panel that highlights “come
back” neighborhoods, with a panel of people from the neighborhood that tell the story of what was
done to strengthen the community.

Maps of neighborhood change at a regional scale can help bring perspective to communities that
had considered themselves immune to affordable housing need. In Seattle, the conversation took a
new turn—
Roosevelt community…is ‘Improve Access’ [type]…[it is one of the] station areas that
were predominantly white, affluent station areas in a wealthier city. When having conver-
sations about what to do with surplus lands the transit agency will have, I was able to go
in and talk about the typology exercise, which highlighted that adopting tools to ensure
affordable housing was a central need for places like Roosevelt… [which] helps counter
some of the community members who want to use those for parks and open space.

The Twin Cities gentrification typology is also meant to educate the suburbs, developers, and oth-
ers who do not comprehend the extent of housing pressure on the urban core. The developer said,
“It’s like driving down the road using your rearview mirror, and all of this demographic change is
in front of you. You’re going to end up in the ditch.”

The maps often serve to validate disenfranchised perspectives. One place where such validation
occurred was Portland, where many planners did not understand the issues—
And then there was this big explosion around a bike lane project…historically black
part of Portland. That was the first wave of displacement. So it’s on the bike boulevards
plan…they were not going to do any of the pedestrian safety stuff that black folks had
asked for. Huge conflicts between bike lanes and buses. So all the transportation planners
were like, ‘Wait what is this gentrification thing people are talking about?’ so that was one
of the first goals was to get people on the same page of what are we talking about.

Cityscape 123
Chapple and Zuk

The maps made the issues more real: “So for the [National] Urban League, and some other black
[organizations, it was like]: ‘See this thing we told you was happening, has happened, is real. It’s in
the data.’”

When the city sponsors the project, as in Portland, it can help legitimize the entire conversation.
The system creator in Portland said, “One of my first conversations with them in talking was, ‘you
should all stop saying that you’re trying to gentrify stuff. It’s not going over well.’ They would routinely
say that. ‘Oh this area needs gentrification…’ with no comprehension of what they were saying…I
think it was really important that there was an acknowledgment on the part of the city that this
was not a purely market accident. So that started happening more in the popular conversation.”

Likewise, a creator of the Chicago maps argues that depicting how neighborhoods are changing,
even where the gentrification process is just barely starting, is effective because users can recognize
themselves—and their own economic struggles—in the maps.
Inequality perpetuates this narrative of gentrification, the fear of gentrification, even if it’s
not really happening. When you can’t get into the middle, when you’re middle income
and you can’t buy a house, then there are structural forces at work. But you want some-
thing to blame, and so the narrative about how gentrification is occurring feels right.

In Chicago, the active dissemination of the index into communities by the University of Illinois at
Chicago researchers helped locals shift into action and policy design. As communities looked at
the new index, they wanted to deconstruct it and shift into figuring out strategies: “Communities
are looking for that sweet spot, where they can prevent excessive development but still get enough
to have resources.” Part of this conversation was spurred by media attention, a radio reporter who
became interested in the issue because of her own neighborhood, Bronzeville. The interest led
the university to add the policy toolkit, which then spurred many new conversations in different
communities.

Another way to use data analysis and maps is to organize or empower a constituency. In Portland,
the housing advocates formed a new coalition and reframed it around displacement, broadly
defined—an umbrella that could include those fighting gentrification, or for renter protections, or
to stabilize communities. For the projects being used by community organizations, all the inter-
viewees reported empowerment and capacity building as outcomes. From Minneapolis-St. Paul to
the San Francisco Bay Area, community organizations use the maps to organize their constituen-
cies. The data do not show only that “it’s real”; the data provide evidence that advocates point to in
meetings with and letters to policymakers.

In St. Louis, the analysis revealed that every rebounding neighborhood had strong civic engage-
ment—
My main surprise is that when we go out to the neighborhoods that we identified as these
rebound neighborhoods, that there really is a—groups on the ground that are talking
about this. About what they can do to help the neighborhood. And it’s sort of, they find it
extremely gratifying to be identified as a neighborhood that’s coming back. There seems
to be a very upbeat conversation about these neighborhoods.

124 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

Interacting with communities about the early warning maps helped creators realize that locals needed
to be equipped to deal with different stages of gentrification. In Chicago: “We quickly realized that
we needed to show people how different tools are appropriate for different stages of gentrification.”
For instance, when gentrification is late stage, as in East Pilsen, the strategy should be to preserve
the diversity and stabilize the community by building coalitions across different groups.

Policy Impact
Many of these early warning systems and reports are in the public domain and have become es-
tablished resources in the ongoing civic conversation about housing. This social context may have
aided the process of policy learning, as policy communities construct shared definitions and debate
ideas (Bennett and Howlett, 1992). Most of the interviewees can point out different ways that the
analyses have shaped the policy conversation, though it is hard to know how pivotal a role they
played in the passage of specific policies. Even the cities using their new neighborhood change
tools internally, such as Los Angeles and Charlotte, reported shifts in how their governments
thought about housing needs and targeting resources.

In St. Louis, the report influenced conversations by the Ferguson Commission about the siting of
Low-income Housing Tax Credit housing in poor areas. Also, the report found that no rebounding
neighborhoods were in the north of the Delmar area, which helped spur a new conversation about
using tax increment financing to fund infrastructure. In Houston, the analysis of gentrification
risk “kept the drumbeat going” at city hall and also helped convince Houston Endowment Inc. to
provide $1 million for a loan fund in a transitioning neighborhood. In Chicago, the maps likely
contributed to the passage of the Single-Room Occupancy Preservation Ordinance. Portland
has a new focus on housing policy throughout the civic arena, with the declaration of a housing
emergency and many different new policies, such as the redistribution of tax increment finance
revenues to affordable housing.

The evidence of policy influence is clearer in the Bay Area, where displacement and gentrification
pressures are particularly acute and the Urban Displacement Project has garnered considerable
media attention (more than 50 articles). Several policymakers responded to the user survey and
said that they use the early warning system to design policy. One local councilmember said, “[I
use the site] to assist in writing public policy for the city I represent as a public official. It is very
valuable and useful.” Another official said, “For my work with the City of Oakland, I used these
to understand how our existing anti-displacement policies could be improved.” San Francisco’s
Mission 2016 Interim Zoning Controls requires developers of new projects in the Mission District
to write a report on their project’s displacement potential, drawing from the early warning system.
City councilmembers in several cities, including San Mateo and San Rafael, California, have refer-
enced the project during council meetings to confirm the city’s displacement risk, show what poli-
cies neighboring municipalities have adopted, and justify passing new antidisplacement policies.
The Berkeley, California mayor used the policy inventory to identify new policies to incorporate
into his comprehensive housing plan. The Metropolitan Transportation Commission is considering
incorporating more stringent antidisplacement targets in its next long-range plan. Affordable hous-
ing producers have used the maps to target sites for subsidized housing development. Unintended
audiences also are using the tools; for example, real estate brokers have reported using the assess-
ment of gentrification risk to identify profitable areas for investment.

Cityscape 125
Chapple and Zuk

It is clear that many users point to the early warning systems to validate their claims that the neigh­
borhood is gentrifying. The tools also seem to be spurring policy changes (though it is impossible
to know whether the new policies would have appeared in the absence of the maps). The method-
ological improvements in the new generation of toolkits have likely helped make policy more effec-
tive. By identifying neighborhoods in early stages of gentrification and displacement, they put the
issue on the radar of local stakeholders; by extending the analysis to the region, the systems clarify
that housing markets operate regionally, affecting peripheral and core areas; and, by incorporating
users into the development of the tools, early warning systems have become more accurate—but
with limitations, as the next section discusses.

Using Early Warning Systems for Prediction

In general, the system developers interviewed did not encourage the use of systems for prediction
but found that policymakers and residents were eager to do so. One expert explained, “Either the
analysis is not very helpful—it is not revolutionary, like predicting change near the metro—or it
is very weak. We can’t predict the [new stadium].” Another pointed to the challenge of accounting
for “sites of reserve,” or property that landlords hold for decades in anticipation of future profit. As
they lay fallow for decades, warning systems may suggest disinvestment, but locals know better.

One interviewee said that developers would generally prefer that the maps be used as a “wake-up call”—
I did not expect that people—especially people in the city—the planning people—to
view it as a predictive model. Or try to keep using it as a predictive model, given that the
whole point was to have very minimal data and simple concepts. So that surprised me.
Was their interest in doing that more so than creating and developing the policy part?
Like, how much more studying of data do you need?

Likewise, in Los Angeles, policymakers describe their tool as a first step. Once they identify areas
that have the potential to change, they can add more qualitative knowledge of the neighborhood in
order to do “prediction”—
We use it as a way to say ‘we can choose between these neighborhoods for the first
[project], and between these for the second one; and then within it, we can focus on a
sub-area, block, commercial corridor, and then we pull in a lot more information. So it’s
definitely not something where you can just enter in some basic search parameter and
then it’ll tell you exactly where to do it… every policy item will have different things to
consider, different political ramifications, and other factors that go well beyond just de-
scribing what’s happened…. And once you have a few good candidates, you have to take
in all these additional considerations. Where are different community groups working in
this particular space that you could partner with?

Even if the developers advise caution, users are eager for more explicit prediction. One expert
explained:
The precise numbers would be valuable in influencing the city. We are now to the point
where we’re hearing the city is ready to have a comprehensive housing plan, and cohesive
housing policy. So precision in numbers would be useful for that. In terms of how we’re
allocating resources.

126 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

Few developers have systematically assessed the validity of their gentrification and displacement
predictions. The exceptions are Houston and the San Francisco Bay Area. Validating their Houston
model using 2007 (2005–2009) American Community Survey data, LISC researchers found 86
percent accuracy for highly susceptible tracts (that is, those that the model predicted were 75 percent
likely to gentrify) and 60 percent accuracy for moderate susceptibility (that is, between 50 and 75
percent likelihood). The Urban Displacement Project found that its analysis from 1990 to 2000
correctly predicted 86 percent of the 85 tracts that gentrified from 2000 to 2013 (Chapple et al.,
2016). The rate of false positives, however, was extremely high: of the 512 tracts that did not
gentrify, the model predicted that 79 percent of the tracts would experience moderate or high gen-
trification. The analysis of household displacement risk revealed the same pattern: a high degree of
accuracy in predicting displacement, but also a high rate of false positives (Chapple et al., 2016).

Both the Chicago and Portland projects used 2010 as the end date for the analysis, so it is possible
to validate those models by checking their results against neighborhood change from 2010 to
2014. Looking at Portland, we found that the extent of vulnerability had changed very little, but
the number of gentrifying or gentrified tracts doubled, from 15 to 30; the model seems to have
underpredicted gentrification, which is occurring very rapidly. In Chicago, we found a near-perfect
correlation (0.94) between the risk score from 2000 to 2010 and that from 2010 to 2014. If anything,
the extent of gentrification has slowed in Chicago; the original analysis found that 11.7 percent
of neighborhoods were gentrifying by 2010, but the 2014 update (using tracts rather than the
original neighborhoods) finds just 8.8 percent.

Given that most developers are skeptical of the accuracy of their own risk assessments, the call of
Pettit and Greene (2016) for better predictive analytics seems warranted. A disinvested neighbor-
hood that receives a false positive “at risk” categorization may resist new market-rate development
or even other forms of revitalization. To the extent that they offer a wake-up call, early warning
systems are helpful for community organizing. Without more precision, however, systems may
actually hinder efforts to develop appropriate policy responses.

Next Steps
Although the first generation of online neighborhood early warning systems has disappeared, a
new set has emerged, now measuring the risk of gentrification and displacement. Policymakers,
community residents, and other stakeholders are actively using these early warning systems strate-
gically, tactically, and for empowerment. Although it is unknown the extent to which the analyses
have actually caused policy shifts, they clearly have influenced the urban debate about housing and
neighborhood change.

The state of predictive analytics is poor, however. Despite methodological advances in the new
generation, the systems are not yet reliable enough to use to design for specific policies. For instance,
they are not able to predict the displacement impacts of specific developments or to identify which
of the many antidisplacement policies is useful in different contexts.

For the most part, the early warning systems studied are not well integrated into the smart cities
movement, potentially missing an opportunity for analytic improvement and long-term sustainability.

Cityscape 127
Chapple and Zuk

None incorporate real-time data on neighborhood change or crowd-sourced data. Unlike the smart
systems that are improving the efficiency of city operations, neighborhood systems have a potential
that is not yet clear—apart from raising awareness and building momentum for policy change. The
smart cities movement has not yet fully grappled with issues of inclusion, instead focusing primar-
ily on efficiency (Pettit and Greene, 2016). One expert said, “Getting the open data movement to
address equity is like moving a big boat.”

The durability of these efforts remains an outstanding question. Of the projects profiled in this
article, a few are planning minor updates, but none have long-term plans to institutionalize this
work. The nonprofit organizations and universities that sponsor much of the work have little
capacity to continue it without a significant influx of resources, and foundation funders come and
go. Although city government is a logical home for early warning systems, especially given new
technological capabilities, the case has yet to be made for why cities should pursue tools and strat-
egies for more equitable, inclusive neighborhood change. Likewise, the private sector has not yet
engaged in neighborhood change debates. Absent such intervention, these early warning systems
will most likely vanish, just as the first generation disappeared.

Acknowledgments
The authors acknowledge comments from two anonymous reviewers and the guest editors and
also from many of the interviewees. They also thank Mitch Crispell for stellar research assistance.
All errors that remain are our own.

Authors
Karen Chapple is a professor in the Department of City and Regional Planning at the University of
California, Berkeley.

Miriam Zuk is Project Director of the Urban Displacement Project at the University of California,
Berkeley.

References
Atkinson, Rowland, Maryann Wulff, Margaret Reynolds, and Angela Spinney. 2011. Gentrification
and Displacement: The Household Impacts of Neighbourhood Change. Melbourne, Australia: Australian
Housing and Urban Research Institute.

Bates, Lisa. 2013. Gentrification and Displacement Study: Implementing an Equitable Inclusive Develop-
ment Strategy in the Context of Gentrification. Portland, OR: Portland State University.

Baud, Isa, Dianne Scott, Karin Pfeffer, John Sydenstricker-Neto, and Eric Denis. 2014. “Digital
and Spatial Knowledge Management in Urban Governance: Emerging Issues in India, Brazil, South
Africa, and Peru (reprinted from vol. 44: 501–509),” Habitat International 44 (October): 501–509.

Bennett, Colin J., and Michael Howlett. 1992. “The Lessons of Learning: Reconciling Theories of
Policy Learning and Policy Change,” Policy Sciences 25 (3): 275–294.

128 Gentrification
 Forewarned: The Use of Neighborhood Early Warning Systems
for Gentrification and Displacement

Chapple, Karen. 2009. Mapping Susceptibility to Gentrification: The Early Warning Toolkit. Berkeley,
CA: Center for Community Innovation.

Chapple, Karen, Paul Waddell, Dan Chatman, Miriam Zuk, Anastasia Loukaitou-Sideris, Paul Ong,
Karolina Gorska, Silvia Jimenez Ramos, and Chhandara Pech. 2016. Developing a New Methodology
for Analyzing Displacement. Draft report. Sacramento, CA: California Air Resources Board.

Craig, William, and Sarah Elwood. 1998. “How and Why Community Groups Use Maps and
Geographic Information,” Cartography and Geographic Information Systems 25 (2): 95–104.

Ding, Lei, Jackelyn Hwang, and Eileen E. Divringi. 2015. Gentrification and Residential Mobility
in Philadelphia. Federal Reserve Bank of Philadelphia Working Paper No. 15-36. https://ssrn.com/
abstract=2675379.

Freeman, Lance. 2005. “Displacement or Succession? Residential Mobility in Gentrifying Neigh-

borhoods,” Urban Affairs Review 40 (4): 463–491.

Galster, George, and Peter Tatian. 2009. “Modeling Housing Appreciation Dynamics in Disadvan-
taged Neighborhoods,” Journal of Planning Education and Research 29 (1): 7–22.

Ghose, Rina. 2011. “Politics and Power in Participation and GIS Use for Community Decision
Making.” In The SAGE Handbook of GIS and Society, edited by Timothy Nyerges, Helen Couclelis,
and Robert McMaster. Thousand Oaks, CA: Sage: 423–438.

Goldsmith, Stephen, and Susan Crawford. 2014. The Responsive City: Engaging Cities through Data-
Smart Governance. New York: Jossey-Bass.

Goodchild, Michael F. 2007. “Citizens as Sensors: The World of Volunteered Geography,” GeoJournal
69 (4): 211–221.

Gorr, Wilpen L., and YongJei Lee. 2015. “Early Warning System for Temporary Crime Hot Spots,”
Journal of Quantitative Criminology 31 (1): 25–47.

Hillier, Amy E., Dennis P. Culhane, Tony E. Smith, and C. Dana Tomlin. 2003. “Predicting Housing
Abandonment With the Philadelphia Neighborhood Information System,” Journal of Urban Affairs
25 (1): 91–106.

Lees, Loretta, Tom Slater, and Elvin K. Wyly. 2008. Gentrification. New York: Routledge/Taylor
and Francis Group.

Maciag, Mike. 2015. “Gentrification in America Report,” Governing the States and Localities Magazine.
http://www.governing.com/gov-data/gentrification-in-cities-governing-report.html.

Marcuse, Peter. 1986. “Abandonment, Gentrification, and Displacement: The Linkages in New
York City.” In Gentrification of the City, edited by Neil Smith and Peter Williams. New York;
Oxford: Routledge: 153–177.

Nathalie P. Voorhees Center. 2015. Gentrification and Neighborhood Change: Helpful Tools for
Communities. Chicago: University of Illinois, Chicago.

Cityscape 129
Chapple and Zuk

———. 2014. The Socioeconomic Change of Chicago’s Community Areas (1970–2010): Gentrification
Index. Chicago: University of Illinois, Chicago.
Oleske, Denise M. 2009. “Health Risks from the Environment: Challenges to Health Service
Delivery.” In Epidemiology and the Delivery of Health Care Services, edited by Denise M. Oleske. New
York: Springer: 211–233.
Pettit, Kathryn, and Solomon Greene. 2016. What If Cities Used Data to Drive Inclusive Neighborhood
Change? Washington, DC: Urban Institute.
Pew Charitable Trusts. 2016. Philadelphia’s Changing Neighborhoods: Gentrification and Other Shifts
Since 2000. Philadelphia: Pew Charitable Trusts. http:// www.pewtrusts.org/philaresearch.
Puget Sound Regional Council (PSRC). 2013. The Growing Transit Communities Strategy. Seattle,
WA: Puget Sound Regional Council.
Ramasubramanian, Laxmi. 2011. “PPGIS Implementation and the Transformation of Planning
Practice.” In The SAGE Handbook of GIS and Society, edited by Timothy Nyerges, Helen Couclelis,
and Robert McMaster. Thousand Oaks, CA: Sage: 411–422.
Sampson, Robert J. 2011. Great American City: Chicago and the Enduring Neighborhood Effect.
Chicago: University of Chicago Press.
Snow, Christopher W., Kathryn L.S. Pettit, and Margery Austin Turner. 2003. Neighborhood Early
Warning Systems: Four Cities’ Experience and Implications for the District of Columbia. Washington,
DC: Urban Institute Metropolitan Housing and Communities Policy Center.
Treuhaft, Sarah. 2006. The Democratization of Data: How the Internet Is Shaping the Work of
Data Intermediaries. Institute of Urban and Regional Development working paper. Berkeley, CA:
University of California, Berkeley.
Turner, Margery, and Christopher Snow. 2001. “Leading Indicators of Gentrification in D.C.
Neighborhoods.” Presentation at the Urban Institute D.C. Policy Forum, Washington, D.C., June
14, 2001. Washington, DC: Urban Institute.
Waddell, Paul. 2002. “UrbanSim: Modeling Urban Development for Land Use, Transportation, and
Environmental Planning,” Journal of the American Planning Association 68 (3): 297–314.
Williams, Sonya, George Galster, and Nandita Verma. 2013. “Home Foreclosures as Early Warning
Indicator of Neighborhood Decline,” Journal of the American Planning Association 79 (3): 201–210.
Wilson, William J. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy.
Chicago: University of Chicago Press.
Winston, Francisca, and Chris Walker. n.d. Predicting Gentrification in Houston’s Low- and
Moderate-Income Neighborhoods. Unpublished paper.
Zuk, Miriam, Ariel H. Bierbaum, Karen Chapple, Karolina Gorska, Anastasia Loukaitou-Sideris,
Paul Ong, and Trevor Thomas. 2015. Gentrification, Displacement, and the Role of Public Investment:
A Literature Review. San Francisco: Federal Reserve Bank of San Francisco. http://www.frbsf.org/
community-development/files/wp2015-05.pdf.

130 Gentrification