ArticlePDF Available

Displacement or Succession?: Residential Mobility in Gentrifying Neighborhoods


Abstract and Figures

This article examines the extent to which gentrification in U.S. neighborhoods is associated with displacement by comparing mobility and displacement in gentrifying neighborhoods with mobility and displacement in similar neighborhoods that did not undergo gentrification. The results suggest that displacement and higher mobility play minor if any roles as forces of change in gentrifying neighborhoods. Demographic change in gentrifying neighborhoods appears to be a consequence of lower rates of intra neighborhood mobility and the relative affluence of in-movers.
Content may be subject to copyright.
Urban Affairs Review
The online version of this article can be found at:
DOI: 10.1177/1078087404273341
2005 40: 463Urban Affairs Review
Lance Freeman
Displacement or Succession? : Residential Mobility in Gentrifying Neighborhoods
Published by:
On behalf of:
The Urban Politics Section, American Political Science Association
can be found at:Urban Affairs ReviewAdditional services and information for Alerts:
What is This?
- Jan 25, 2005Version of Record >>
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
10.1177/1078087404273341 URBAN AFFAIRS REVIEW / March 2005Freeman / RESIDENTIAL MOBILITY
Residential Mobility in
Gentrifying Neighborhoods
Columbia University
This article examines the extent to which gentrification in U.S. neighborhoods is associated with
displacement by comparing mobility and displacement in gentrifying neighborhoods with mo
bility and displacement in similar neighborhoods that did not undergo gentrification. The results
suggest that displacement and higher mobility play minor if any roles as forces of change in gen
trifying neighborhoods. Demographic change in gentrifying neighborhoods appears to be a con
sequence of lower rates of intraneighborhoodmobility and the relative affluence of in-movers.
Keywords: gentrification; neighborhood change; displacement; housing mobility
Gentrification, the process by which decline and disinvestments in inner-city
neighborhoods are reversed, has emerged as one of the most controversial
issues in the urban United States today. By attracting middle-class residents
and spurring investment, gentrification has the potential to revitalize depres
sed central city neighborhoods. After decades of disinvestment, middle-class
exodus, and declining tax bases, some view this as a welcome development.
The threat of displacement, however, whereby current residents are forced
to move because they can no longer afford to reside in the gentrifying neigh
borhoods has become such a concern that some are reflexively opposed to
gentrification. Indeed, the fear of displacement has been one of the motivat
ing forces behind community activists organizing against gentrification.
Moreover, some scholars look askance at gentrification on the assumption
that it harms the indigenous residents by displacing them (Hartman 1979;
Smith 1996).
URBAN AFFAIRS REVIEW, Vol. 40, No. 4, March 2005 463-491
DOI: 10.1177/1078087404273341
© 2005 Sage Publications
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
When we look for empirical evidence of displacement, the process by
which neighborhoods putatively undergo gentrification, the evidence is
much less definitive. Although displacement has become synonymous with
gentrification in the way that White flight has become synonymous with
racial transition, unlike the latter type of neighborhood change (see Crowder
2000), there is relatively little in the way of persuasive empirical evidence
that suggests this is indeed how gentrifying neighborhoods change.
This study aims to shed light on the process of neighborhood change
in gentrifying neighborhoods. More specifically, it will test whether dis
placement is the driving force behind demographic change in gentrifying
As an unanticipated phenomenon that contradicted the prevailing wisdom
on urban decline, much of the early scholarly work on gentrification focused
on theorizing about the origins, causes, and meaning of gentrification in
the postindustrial urban United States. Scholars first sought to document
whether the phenomenon of inner-city reversal or rebirth was actually occur-
ring and, if so, to what extent (Baldassare 1982; Clay 1979; James 1977;
Lipton 1977; National Urban Coalition 1978; Sumka 1979). These studies
were consistent in showing that although small in the overall scheme of met-
ropolitan changes, gentrification was indeed a reality in many communities
in older central cities during the 1970s. Having documented gentrification’s
existence, attention turned to theorizing about its origins and meanings for
cities. Theorists debated about the importance of changing demographics,
changing tastes, and the “rent gap” that made inner-city investing profitable.
What emerged from this debate was recognition of the importance of chang
ing demographics, tastes, and professional services clustering in the cities to
provide the gentrifiers, and a history of disinvestment that created ripe oppor
tunities for reinvestment in certain neighborhoods as preconditions for gen
trification (Beauregard 1986; Hamnett 1991; Ley 1980; Rose 1984; Smith
Along with theorizing about the causes of gentrification, concern about
displacement emerged contemporaneously. Anecdotal reports of displace
ment and the demographic changes that were obviously taking place in
gentrifying neighborhoods led many to believe that displacement was a
widespread phenomenon and the engine behind demographic change in gen
trifying neighborhoods (Hartman 1979).
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
When social scientists attempted to empirically quantify displacement in
gentrifying neighborhoods, however, their findings typically failed to find
evidence of widespread displacement (Sumka 1979). The first wave of dis
placement studies generally used two types of methods to measure displace
ment due to gentrification. One was a succession methodology that examines
how the characteristics of in-movers differ from those of out-movers (Henig
1980; Spain, Reid, and Long 1980). The other common approach was to ret
rospectively ask people why they moved from their former residence (Grier
and Grier 1978; Newman and Owen 1982; Lee and Hodge 1984).
Succession studies can only help to define the upper bound of displace
ment, however; they cannot be used to determine whether housing or neigh
borhood transitions occurred through the induced departure of low-income
households or through normal housing turnover and succession, because
they do not consider other reasons why households might move besides dis
placement. Succession studies can, thus, verify that the process of gentrifica
tion is underway but, without additional information, cannot demonstrate its
impact on displacement.
Displacement studies based on asking respondents why they moved gen-
erally define the displaced as those who are forced to move for reasons that
are beyond the household’s control and related to conditions in the dwelling
or the surrounding neighborhood (Grier and Grier 1978). The biggest prob-
lem with studies that focus on retrospective motives for moving is that they
typically fail to identify the location of the respondent’s former residence.
Consequently, it is impossible to determine how much, if any, of the displace-
ment observed is due to gentrification as opposed to other environmental fac-
tors. Lee and Hodge (1984) also attempted to focus on displacement that
could be attributed to gentrification by comparing metropolitan versus
nonmetropolitan, northeastern/midwestern versus the rest of the country, and
central city versus suburban displacement rates. They hypothesized that gen
trification would be highest in central cities in the Northeast/Midwest,
and consequently, displacement would be greatest there also. To test their
hypothesis, they used data from the American Housing Survey.
While Freeman and Braconi’s (2004) and Vigdor’s (2002) results chal
lenge some common conceptions about the neighborhood dynamics behind
gentrification their results can hardly be viewed as the final word on displace
ment. For the most part, their hypotheses were not borne out. As a percentage
of all movers, the overall displacement rate was 3.31%. The highest rate they
observed was 8.9% in Kansas City, and the lowest, 1%, was in suburban
Cincinnati. Yet because their definition of displacement includes forces not
plausibly linked to gentrification, such as abandonment and mortgage de
fault, one can’t be certain if a narrower definition of displacement would have
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
proved their hypothesis correct. Moreover, their level of geographic aggre
gation may have been far too great to detect disturbances occurring at a very
localized level.
Aside from failing to quantify displacement due to gentrification in a con
vincing fashion, these early displacement studies also failed to shed much
light on what happened to the putative displacees. Did they stay in the neigh
borhood? Critical questions such as this remained unanswered despite the
efforts of early gentrification scholars.
Schill and Nathan (1983) attempted to address these questions by focus
ing on gentrifying neighborhoods and the individuals moving out of them by
using a narrow definition of displacement that could be directly attributable
to gentrification. The principal drawback to Schill and Nathan’s method was
that no baseline displacement rate could be estimated. Consequently, one
cannot compare displacement rates in gentrifying and nongentrifying areas.
Moreover, there is no measure of the relative mobility of households in dif
ferent types of neighborhoods, so a higher percentage of moves from gen-
trifying areas may be displacements, whereas the aggregate number of
displacements from those neighborhoods may be the same or lower. To deter-
mine whether gentrification causes an increased number of poor households
to be displaced, there must be a basis of comparison to other neighborhoods
in which gentrification is not occurring.
This methodological flaw in the Schill and Nathan (1983) study is typical
of the first wave of displacement studies. Missing from most of the earlier
analyses was some type of counterfactual that would inform us how much
displacement would have occurred absent gentrification. A common wisdom
nevertheless emerged that gentrification affected preexisting residents pri
marily by displacing them.
Two recent studies on gentrification and displacement in the United
States, however, cast suspicion upon the common wisdom. These studies
found that poor households and those without college degrees who resided in
gentrifying neighborhoods were less likely to move than similar households
residing elsewhere (Freeman and Braconi 2004; Vigdor 2002).
These studies were limited to Boston and New York City, which are in
some ways atypical American cities. Moreover, both of these studies used
geographic entities that are much larger than what is typically thought of as a
neighborhood. Freeman and Braconi used subborough areas in New York
City, which comprised an average of 131,000 persons in 1999. Likewise, the
zones Vigdor used as proxies for neighborhoods had between 100,000 and
200,000 residents. Quantitative studies typically use much smaller levels of
geography, such as a census tract with about 4,000 people, to define a neigh
borhood. Although the larger geographic areas used in the Freeman and
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
Braconi and Vigdor studies were informative, the results would be more
convincing if smaller level of geographies were used.
Finally, criticisms can be raised about the counterfactuals Freeman and
Braconi (2004) and Vigdor (2002) used in their studies. By comparing
mobility rates in gentrifying neighborhoods to other neighborhoods, these
studies certainly were an improvement over previous studies, which typically
included no baseline or comparison group. Nonetheless, it is questionable
that all other neighborhoods are the appropriate “control” group. More
appropriate would be neighborhoods that were similar to gentrifying neigh
borhoods except for undergoing gentrification. These neighborhoods come
closer to being controls in the sense that they are similar to the gentrify
ing neighborhoods but did not receive the “treatment” that in this case is
This research corrects for these limitations in the following ways: A
national sample is used, thus extending the external validity of the results;
census tracts, which come much closer to approximating what is typically
thought of as a neighborhood, are the geographic entities used; mobility in
gentrifying neighborhoods is compared to mobility in potentially gentrifying
neighborhoods that did not gentrify (hereafter referred to as nongentrifying);
and some analysis of the destination neighborhoods of movers in gentrifying
neighborhoods is provided. The next section explains the methods used to
accomplish these improvements.
Data for this study are drawn from the geocoded version of the Panel
Study of Income Dynamics (PSID). The geocodes were used to link the PSID
to tract and metropolitan level data from the decennial census. The PSID is a
longitudinal survey of a representative sample of U.S. individuals and the
families with which they reside (Hill 1992). The PSID has been following the
same individuals and their families since 1968. The geocoded version of the
PSID is an especially powerful tool for examining how households respond
to gentrification. The PSID collects a plethora of social and economic infor
mation, and by following the same households throughout time, one can
observe how their status changes in relation to the characteristics of the
neighborhood, including gentrification. Moreover, we can observe what hap
pens to preexisting residents who move out of gentrifying neighborhoods.
The 1986-1999 sample years will be included in the analysis. The PSID only
began collecting important housing information in 1986, hence the rationale
for beginning the analysis in 1986. Each respondent will be matched with the
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
census geographies and data for the corresponding years. Census data and
geographies for 1980 will be assigned to the 1986-1989 PSID sample years,
and census data and geographies for 1990 will be assigned to the 1990-1998
PSID sample years.
The geocoded version of the PSID allows researchers to link each respon
dent in the sample to the specific census tract or other census geographies that
they reside in. The decennial census is the most comprehensive source of
neighborhood-level data. As shown below, this data can be used to identify
gentrifying neighborhoods or tracts.
The unit of analysis will be individuals who are identified as household
heads in the PSID sample. The key independent variables will be residence in
a gentrifying neighborhood. The sample will be limited to household heads
to preclude counting moves or changes in economic status by members of the
same household more than once. Because gentrification is by definition an
urban phenomenon, the analysis is further limited to households residing in
metropolitan areas.
To make use of the longitudinal nature of the PSID, the data will be struc-
tured in a “person-year” format in which each observation represents the
characteristics of an individual and that individual’s environment in a spe-
cific year. Each household head in the PSID will thus contribute x number of
records to the analysis data set with x representing the number of years they
are household heads. Individual-level covariates are allowed to vary through
out time. A discrete time logistic regression model will be used to model how
the likelihood of moving or being displaced is affected by residence in a gen
trifying neighborhood. Using the discrete time approach allows the influence
of length of time in residence to be modeled explicitly. The regression model
allows other potentially confounding factors, described below, to be held
Residential mobility and displacement, respectively, represent the de
pendent variables of interest in the multivariate analyses. Displacement will
be proxied for using two measures. The first measure uses all types of resi
dential mobility as a proxy for displacement. The rationale behind this
approach is that to the extent we can model residential mobility in general
using the life-cycle theory of residential mobility, any surfeit of mobility in
gentrifying neighborhoods as compared to other neighborhoods can be
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
attributed to gentrification-induced displacement. Residential mobility be
tween time t and time t + 1 will therefore be the dependent variable here.
The second proxy will consider as displaced all those respondents in the
PSID sample who give as their reason for moving in the previous year that
they wanted to consume less space, wanted to pay less rent, or moved in
response to outside events including being evicted, health reasons, divorce,
joining the armed services, or other involuntary reasons. Although this cate
gory includes some responses that might not be considered displacement, the
PSID categorization of responses precludes separating them out. This mea
sure will overstate the extent of displacement due to gentrification, because
this type of displacement is usually conceptualized as households moving
because they were evicted or because their housing costs became prohibitive
due to rising housing costs in their neighborhood. Nevertheless, this defini
tion will reveal an upper bound on the extent to which displacement appears
to be caused by gentrification. Thus, all household heads who moved be
tween time t and time t+1for the reasons described above will be considered
The key independent variable of interest is residence in a gentrifying
neighborhood. It is the impact of such residence on displacement and resi-
dential mobility that this research seeks to measure. The measurement of
gentrification, however, has been subject to debate. The Encyclopedia of
Housing (Smith 1998, 198) defines gentrification as “the process by which
central urban neighborhoods that have undergone disinvestments and eco
nomic decline experience a reversal, reinvestment, and the in-migration of a
relatively well-off, middle- and upper middle-class population. Hammel
and Wyly (1996, 250) define gentrification as “the replacement of low-
income, inner-city working class residents by middle- or upper-class house
holds, either through the market for existing housing or demolition to make
way for new upscale housing construction.” The U.S. Department of Hous
ing and Urban Development defined gentrification as the process by which a
neighborhood occupied by lower-income households undergoes revitaliza
tion or reinvestment through the arrival of upper-income households” (U.S.
Department of Housing and Urban Development 1979, 4).
Although there is no one consensual definition of gentrification, certain
dimensions appear consistently among the different definitions. First, con
sider the types of neighborhoods with the potential to be gentrified. Charac
teristics of such neighborhoods would include (1) central city neighborhoods
(2) populated by low-income households that have previously experienced
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
(3) disinvestment. Next, consider the actual process of gentrification. The
definitions listed above point to an (4) influx of the relatively affluent or gen
try, and (5) an increase in investment. Accordingly, this research will attempt
to capture these five dimensions in operationalizing gentrification. The first
three represent disadvantaged neighborhoods that are the pool of potentially
gentrifying neighborhoods, whereas the last two refer to the process of gen
trification. Because gentrification is a dynamic process, it is necessaryto
compare changes in tract characteristics throughout time to determine which
ones appear to be gentrifying. Comparisons will be made for the intercensal
periods 1980-1990 and 1990-2000. Although one could argue for including
suburban neighborhoods, the definitions above and the literature clearly sug
gest that gentrification is primarily a central city phenomenon, hence the
decision to exclude suburban neighborhoods.
Several of the dimensions described in definitions of gentrification havea
relative component to them. For example, the notion that low-income house
holds originally populated gentrifying neighborhoods begs the question of
incomes lower than what? Lower relative to the rest of the country? Lower
relative to that central city? Most analysts would agree that the metropolitan
area comes closest to approximating the organic regions that we would con-
sider housing or labor markets. Consequently, the metropolitan area within
which a particular neighborhood is embedded will serve as the reference area.
Census tracts identified as part of central cities by the U.S. Department of
Housing and Urban Development were used to designate central city neigh-
To capture the dimension of low-income households, neighbor-
hoods with median incomes at or less than the median in their respective met-
ropolitan areas will be included. Disinvestment is a perhaps more difficult
concept to measure using census data, because the census collects no infor
mation on investment activity. One can try to proxy for disinvestments by
considering how much of the housing stock is of recent vintage. Neighbor
hoods where a substantial portion of the housing stock was built in recent
years would not seem to be prime candidates for gentrification. Following
this logic, we compare the proportion of housing built within the past 20
years to the metropolitan-wide average. Those neighborhoods with the pro
portion of their housing stock built within the past 20 years falling below the
median for their respective metropolitan areas are considered candidates for
gentrification. The median is an admittedly arbitrary threshold for both of
these indicators. But the definitions of gentrification described above suggest
that potentially gentrifying neighborhoods are those in the lower end of the
socioeconomic distribution. To check the robustness of the results, the 30th
and 40th percentile, respectively, were also used. The 30th percentile, how
ever, produced too few tracts that could be matched to the PSID sample. The
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
40th percentile threshold results are illustrated along with the results for
the median defined gentrifying neighborhoods. Additional specifications of
gentrification were attempted as described in the sensitivity analysis section
Next, consider the process of gentrification. One of the dimensions of this
process is an influx of the “gentry” or relatively affluent households. Besides
income, however, the notion of gentrification has more of a connotation of
class. For the purposes of this study, education is perhaps a better marker of
class than income. Income fluctuates throughout time, whereas among adults
educational attainment levels are relatively stable. Moreover, young artists
and professionals who have relatively low incomes often pioneer gentrifica
tion. A measure of gentrification relying on income might overlook neigh
borhoods experiencing an influx of highly educated but poorly paid profes
sionals, whereas a measure based on education would be less likely to miss
this type of change. Using education as opposed to income also helps to dis
tinguish between incumbent upgrading among current residents as opposed
to gentrification fueled by outsiders (Clay 1979). Consequently, to capture
the implied change in class associated with gentrification, changes in educa-
tional attainment are considered.
There has been, however, a societal-wide increase in educational attain-
ment (U.S. Census Bureau 2003). Therefore, an increase in educational
attainment at the neighborhood level might simply reflect the universal trend
in increased educational attainment. What we are interested in is an increase
in educational attainment at least equal to the overall trend. The increase in
educational attainment in intercensal years is therefore contrasted to the
increase in educational attainment in the respective metropolitan areas.A
neighborhood must have an increase in educational attainment, measured as
the percentage of those 25 years and older with at least four years of college,
greater than or equal to the average increase in educational attainment in the
neighborhoods respective metropolitan area.
The final criterion used to distinguish gentrifying from other neighbor
hoods is reinvestment. As mentioned earlier, the decennial census has no
information on investment, but housing prices can be used to proxy for
investments. An increase in housing prices in a neighborhood will therefore
be considered as the final criterion for gentrification. Thus, a neighborhood
must meet the following criteria to be considered gentrifying:
1. Be located in the central city at the beginning of the intercensal period.
2. Have a median income less than the median (40th percentile) for that metro
politan area at the beginning of the intercensal period.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
3. Have a proportion of housing built within the past 20 years lower than the pro
portion found at the median (40th percentile) for the respective metropolitan
4. Have a percentage increase in educational attainment greater than the median
increase in educational attainment for that metropolitan area.
5. Have an increase in real housing prices during the intercensal period.
Neighborhoods meeting the first three criteria can be thought of as poten
tially gentrifying. Those that meet the first three criteria but not the last two
are termed nongentrifying. Finally, neighborhoods meeting all five criteria
will be considered gentrifying. We therefore have two definitions of gentrifi
cation hereafter referred to as median defined and 40th percentile defined.
At this point, it is worthwhile to consider whether the neighborhoods
identified using the above criteria would typically be thought of as gentrify
ing. Table 1 shows how the gentrifying and nongentrifying neighborhoods
changed between 1980 and 1990, and Table 2 shows how these neighbor
hoods changed between 1990 and 2000.
Both of these tables are based on
categorizations of potentially gentrifying neighborhoods using the median.
Categorizations using the 40th percentile are available from the author upon
Comparisons made here are between gentrifying neighborhoods and
nongentrifying neighborhoods. Across both decades, housing prices clearly
increased more steeply in the neighborhoods classified as gentrifying, espe-
cially owner-occupied housing. Likewise, levels of educational attainment
grew more steeply in those neighborhoods classified as gentrifying. This
should come as no surprise, because these two dimensions were used to de-
fine gentrifying neighborhoods. The poverty rate also declined more quickly
in neighborhoods classified as gentrifying, consistent with what might be
expected in neighborhoods on the upswing. Both neighborhoods classified
as gentrifying and nongentrifying experienced declines in the proportion
White in their neighborhoods, but the declines were greater in nongentri
fying neighborhoods. Again, to the extent that the gentry are more likely to
be White, this pattern is consistent with what might be expected in gentrify
ing neighborhoods. When we turn to median household income (family
income in 1980), the results are less consistent with what might be expected.
Between 1980 and 1990, household income increased more rapidly in gen
trifying neighborhoods, but the pattern was reversed between 1990 and
2000. This could be because household sizes were becoming smaller in gen
trifying neighborhoods between 1990 and 2000, and/or educated but rela
tively poor pioneers were responsible for much of the gentrification in the
1990s. Overall, however, the pattern of neighborhood change depicted
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
TABLE 1: Characteristics of Neighborhoods, 1980-1990 Median
Gentrifying Tracts Nongentrifying Tracts All Other Urban Tracts
1980 1990 1980 1990 1980 1990
Housing price $52,676 $119,841 $52,154 $77,096 $69,888 $108,803
Home ownership rate (%) 50 46.5 49.8 43.6 63 58.5
College graduate (%) 11.7 18.1 15.1 15.7 17.5 22.1
Median household income $28,366 $38,873 $27,208 $32,731 $50,750 $43,984
Poverty rate (%) 15.8 15.1 16.1 17.8 12.1 12.5
White (%) 69.2 63.7 68.1 61.1 77.2 73.4
Rent $411 $502 $356 $198 $567 $358
N 2,336 3,338 34,716
NOTE: Dollar figures are adjusted for inflation.
Characteristics of Neighborhoods, 1990-2000 Median
Gentrifying Tracts Nongentrifying Tracts All Other Urban Tracts
1990 2000 1990 2000 1990 2000
Housing price $74,823 $113,253 $75,135 $96,004 $123,661 $159,819
Home ownership rate (%) 41.3 46.2 39.4 43.1 61.8 66.6
College graduate (%) 12.1 19.3 14.5 15.4 23.6 27.6
Median household income $34,413 $30,419 $27,279 $28,300 $48,923 $50,534
Poverty rate (%) 25.1 23.1 24.2 25.7 9.6 10.4
White (%) 54.7 48.9 48.2 40.0 78.2 71.3
Median rent $489 $450 $526 $444 $703 $620
N 2,808 6,201 28,237
NOTE: Dollar figures are adjusted for inflation.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
in Tables 1 and 2 is consistent with what might be expected in gentrifying
The central question posed here is how gentrification affects the displace
ment/mobility of preexisting residents. This of course implies a counter
factual, that is, how would preexisting residents have fared if their neighbor
hoods did not undergo gentrification? Thus, two comparisons are made: to
nongentrifying neighborhoods and to all other urban neighborhoods. The
key comparison is between gentrifying neighborhoods and nongentrifying
neighborhoods. This comparison will tell us if displacement/mobility is
higher in gentrifying neighborhoods, as would be expected to the extent that
gentrification causes displacement. Nongentrifying neighborhoods include
those neighborhoods that were located in the central city at the beginning of
the intercensal period, had a median income less than the median (40th per
centile) for that metropolitan area at the beginning of the intercensal period,
and had a proportion of housing built within the past 20 years lower than the
proportion found at the median (40th percentile) for the respective metro-
politan area. Thus, the independent variable measuring residence in a gen-
trifying neighborhood will be captured by three categories: (1) preexisting
household heads in gentrifying neighborhoods, (2) households heads in non-
gentrifying neighborhoods, and (3) household heads in all other metropoli-
tan neighborhoods.
Despite the correlational validity demonstrated by illustrating how neigh-
borhoods classified as gentrifying changed, the measure described above is
unlikely to perfectly capture the phenomenon of gentrification. In recogni-
tion of this likely shortcoming, I also use the rate of rental inflation as an inde-
pendent variable. Rental inflation, perhaps even more than increases in prop
erty values, is the causal mechanism often pointed to as responsible for
displacement in gentrifying neighborhoods. After all, homeowners are some
what protected against increases in housing prices. In contrast, as a neighbor
hood gentrifies, landlords will be able to command a higher rent for their
units and will raise prices accordingly. Renters whose income may or may
not be rising concomitantly with rental increases may be especially vulnera
ble to displacement under these circumstances.
Rental inflation will be measured as the percentage increase in rental
prices between 1980 and 1990 for all the years 1986-1989 and the increase in
rental prices between 1990 and 2000 for all the years 1990-1997. Rental
prices will be measured using data from the 1980, 1990, and 2000 censuses.
To discern the impact of gentrification on displacement, a key assumption
that must be met is that the households in gentrifying neighborhoods are sim
ilar to households elsewhere. Because households select the neighborhoods
they reside in, characteristics associated with their choosing gentrifying
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
neighborhoods might also affect how they respond to gentrification. Never
theless, a first step toward understanding how gentrification might affect pre
existing households is to observe whether gentrification is correlated with
outcomes such as displacement. Moreover, by controlling for other factors
associated with these outcomes, our confidence in the observed relationship
between gentrification and these outcomes will be enhanced. Outlined below
are the theoretical underpinnings and conceptualization of the factors tobe
held constant while discerning the relationship between gentrification and
each of the respective aforementioned outcomes.
The literature on residential mobility suggests that households move
when there is a discrepancy between their current housing needs and their
current housing unit. Life cycle factors are the prime catalysts of these dis
crepancies and hence should be controlled for (Rossi 1980; Speare 1974).
Life cycle factors refer to major life events, such as taking a new job, getting
married, or having a child, which are likely to trigger a change in one’s hous
ing needs and necessitate moving. An individual’s age, gender, marital sta-
tus, and parental status will serve as proxies for the life cycle and will be
included as control variables in the analysis.
In addition to life cycle factors, housing conditions are also likely to influ-
ence the likelihood of someone moving. The availability of other housing
opportunities should influence the likelihood of someone moving. The va-
cancy rate in the metropolitan area is therefore included as a measure of other
housing opportunities. Moreover, homeowners are less likely to move be-
cause of the higher transactions costs (i.e., finding a buyer) that homeowners
face. In addition, homeowners are likely to face less displacement pressure
from gentrification. Residents of subsidized housing are also less likelyto
face such displacement pressure. Consequently, whether someone is a home
owner or a recipient of subsidized housing is included as a control in the
multivariate analyses.
The final housing condition held constant will be
whether a household is crowded, which is defined using the convention of
having more than one person per room.
The longer one lives somewhere, the stronger that person’s ties are likely
to be to the surrounding area and the less likely he or she is to move. The
length of time a household head has been residing at his or her current resi
dence is therefore included as a predictor of residential mobility.
Employment opportunities are also a prime motivator of residential mo
bility. Someone having difficulty securing employment might be expected to
be more likely to move. To account for this, the amount of time the household
head was unemployed in the previous week is included in the analysis.
Prior research on residential mobility has also shown income and house
hold size to be related to the likelihood of moving (Fielding 1994). House
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
hold size and income are therefore also controlled for in the analysis. Ad
ditional controls included in the analysis are race/ethnicity, region of the
country, and the year of the move.
Certain households are more likely than others to be affected by the
changes wrought by gentrification. For example, poor renter households
would seem to be especially vulnerable to potential displacement from gen
trification, for it is the poor who may be least able to afford concomitant
increases in housing costs. Likewise, renters would probably be more sus
ceptible to displacement pressure, because they have much less control over
their unit than owners. This discussion suggests that an effort be made to dis
cern whether gentrification affects especially vulnerable households differ
ently from other households. Whether a household is a poor renter is the mea
sure of vulnerability that will be taken into account here. To determine if
gentrification affects poor renters differently from other households, interac-
tion terms between the independent variables representing residence in a
gentrifying neighborhood or the level of rental inflation in the respondents
neighborhood and an indicator of whether a household is a poor renter are
included in the models. A positive and statistically significant interaction
term would indicate that gentrification had a greater impact, meaning that
poor renter households were more likely to be displaced or move than other
households in gentrifying neighborhoods (Jaccard 2001). Table 3 illustrates
the means of the variable used in the analysis.
1. Preexisting residents of gentrifying neighborhoods are more likely to move/be
displaced when residing in gentrifying neighborhoods, all things being equal.
2. Poor renters residing in gentrifying neighborhoods are more likely to move/be
displaced when residing in gentrifying neighborhoods, all things being equal.
To summarize, the analytic strategy used here is to compare rates of displace
ment/mobility between household heads in gentrifying neighborhoods and
household heads residing in neighborhoods that were otherwise similar at the
beginning of the decade but did not gentrify. To complement this, the rela
tionship between rental inflation and displacement/mobility is also dis
cerned. Finally, for all the displacement/mobility analyses, attempts are
made to discern if poor renters are especially vulnerable to displacement or
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
Table 4 illustrates the bivariate relationships between gentrification and
displacement/mobility. The relationships here and in subsequent tables are
Descriptive Statistics
Variable Frequency
Moved 11.7%
Displaced 1.3%
Gentrifying neighborhood (median definition) 6.5%
Nonpotentially gentrifying neighborhood (median definition) 43.4%
Gentrifying Neighborhood (40th percentile definition) 3.9%
Nonpotentially gentrifying neighborhood (40th percentile definition) 46.7%
Poor 19.2%
No college 51.4%
Renter 23.7%
Transitory income $60,695
Permanent income $43,865
Years in residence 7.6
Black 11.6%
Other 0.5%
Latino 1.6%
Age 25-34 29.8%
Age 35-44 25.9%
Age 45-54 17.7%
Age 55-64 14.8%
Older than 65 17.6%
Female 28.3%
Married 59.3%
Divorced, separated, or widowed 26.5%
Has children 38.1%
Family size 2.6
Weeks unemployed last year 0.8
Immigrant 1.6%
Unit is crowded 2.3%
Resides in subsidized unit 3.7%
Vacancy rate 2nd quintile 25.5%
Vacancy rate 3rd quintile 13.1%
Vacancy rate 4th quintile 19.5%
South 29.8%
Midwest 28.7%
West 18.9%
Year between 1990 and 1997 69.4%
Sample size 45,108
person years
a. Person years means that each record in the data represents 1 year for a particular individual.An
individual may contribute more than 1 year to the data.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
expressed as odds ratios. Values greater than 1 indicate a positive relationship
between the independent or control variable and displacement and mobility
whereas values of less than 1 indicate a negative relationship. The second and
third columns of Table 4 suggest gentrification as defined by the criteria on p.
10 is not related to displacement/mobility. Rental inflation, however, is.
Whether this these bivariate relationships will persist when statistical con-
trols are added is addressed in Table 5.
The results of the statistical models described above are presented in two
ways in Tables 5 and 6, below. The top rows illustrate the relationship
between the independent and control variables and displacement/mobility.
The bottom rows of Tables 5 and 6 provide the predicted probabilities of a
household moving or being displaced in a gentrifying neighborhood and a
nongentrifying neighborhood. These predicted probabilities are calculated
by setting the independent and control variables to their respective mean val
ues or, in the case of categorical variables, their proportions in the sample.
The interaction terms that tested whether poor renters were especially sus
ceptible to displacement/mobility proved to be insignificant and hence were
dropped from the final models, but are illustrated in abbreviated form in
Table A1 (in the appendix). The lack of significant interaction terms would
seem to suggest that poor renters were not more susceptible to displacement.
The counterintuitiveness of this result, however, makes one hesitant to rule
out the possibility that poor renters are indeed more susceptible to displace
ment but that these models were not able to detect such an effect.
The multivariate analyses with measures of gentrification as the inde
pendent variable are presented in Table 5. The second and third columns of
Table 5 show the results of modeling the likelihood of moving. These results
are not suggestive of a relationship between mobility and gentrification. Both
Bivariate Relationship Between Mobility and Gentrification/Rental
40 Percentile Median Rental
Threshold Threshold Inflation
Moved in Past Year as Dependent Variable
Gentrifying neighborhood 0.73 0.90
Other neighborhood 1.31** 0.91
Displaced as dependent variable 1.46 0.79
1.63 0.88
**p < .05. ***p < .001.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
Relationship Between Gentrification and Mobility/Displacement
Moved Displaced
Median 40th Percentile Median 40th Percentile
Gentrifying neighborhood
0.96 0.80 1.62*** 1.90***
Other neighborhood
0.87 0.89 2.28* 2.66**
Poor 0.93 0.93 1.23 1.21
Renter 1.90*** 1.90*** 3.65*** 3.63***
Resides in assisted housing 0.79 0.79 0.63* 0.65
Transitory income 0.99 0.99 1.01** 1.01**
Years in residence 1.01 1.01 0.98*** 0.98***
Black 0.83 0.84 1.08 1.06
Other 0.61* 0.62* D D
Latino 1.20* 1.21* 1.60 1.59
Age 25-34 0.69 0.69 0.83 0.81
Age 35-44 0.65** 0.64** 0.77 0.74*
Age 45-54 0.40*** 0.39*** 0.21*** 0.21***
Age 55-64 0.37*** 0.36*** 0.41** 0.42**
Older than 65 0.35*** 0.35*** 0.41** 0.40**
Female 0.82** 0.82** 0.80 0.81
Married 0.99 0.99 0.42*** 0.42***
Divorced, separated, or widowed 0.94 0.94 0.76 0.76
Has children 1.08 1.08 0.67 0.68
Family size 0.84*** 0.84*** 1.14* 1.14*
Permanent income 1.01*** 1.01*** 0.99*** 0.99***
Weeks unemployed last year 1.01 1.01 1.02*** 1.02***
Immigrant 0.92 0.91 0.78 0.78
High school graduate 0.89 0.88 1.08 1.09
Some college 0.92 0.91 0.80 0.80
College graduate 0.90 0.88** 0.66 0.66
Unit is crowded 1.52 1.54 1.48 1.47
Vacancy rate 2nd quintile 1.17 1.17 1.40* 1.18
Vacancy rate 3rd quintile 1.34*** 1.34** 1.17 1.79**
Vacancy rate 4th quintile 1.69*** 1.69** 1.79*** 1.60
South 1.55*** 1.55*** 1.01 .98
Midwest 1.20* 1.20* 1.12 1.12
West 1.04 1.04 2.14*** 2.15***
Year between 1990 and 1997 0.16*** 0.16*** 0.82 0.78*
F statistic 1,335.55*** 1,403.45*** 764.97*** 867.88***
N 31,547
Probability of Moving Probability of Being Displaced
Gentrifying neighborhood 0.14 0.10 0.013** 0.014**
Nongentrifying neighborhood 0.14 0.13 0.009** 0.009**
NOTE: D = dropped due to multicollinearity.
a. The reference category here is nongentrifying neighborhoods.
*p < .01. **p < .05. ***p < .001.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
the 40th percentile and median defined gentrifying neighborhoods have neg
ative, albeit statistically insignificant, relationships with gentrification. The
predicted probabilities illustrate the likelihood of a household moving if re
siding in gentrifying or nongentrifying neighborhoods. The predicted proba
bilities are not higher in the gentrifying neighborhoods. Thus, both the odds
ratios and predicted probabilities are inconsistent with gentrification being
associated with displacement.
The fourth and fifth columns show the results of modeling the likelihood
of being displaced. Here, there are positive and statistically significant rela
tionships between residing in a gentrifying neighborhood and displacement.
It should also be noted that when compared to nongentrifying neighbor
hoods, displacement rates are higher in other neighborhoods as well. Appar
ently, it is in the nongentrifying neighborhoods where the threat of displace
ment is lowest.
To get a better sense of the substantive meaning of the relationship be
tween displacement and gentrification, consider the predicted probabilities
for displacement for those living in gentrifying neighborhoods and those liv-
ing in nongentrifying neighborhoods. The bottom rows of Table 5 show that
the probability of being displaced ranges from 0.9% to 1.4%. The predic-
tions suggest that the incremental increase in the probability of displace-
ment as a result of gentrification is small, perhaps in part because displace-
ment is a relatively rare occurrence regardless of what type of neighborhood
one resides in. The probability of being displaced in a gentrifying neigh-
borhood is about 0.5% higher than in a nongentrifying neighborhood. It
should also be recalled that this definition of displacement includes moves
for health reasons, divorce, joining the armed services, or other involuntary
reasons and thus probably overstates the amount of gentrification-induced
Overall, the models suggest at most a modest link between gentrification
and displacement. The relationship between mobility and gentrification is
not statistically significant. Although displacement was significantly related
to gentrification, the substantive size of this relationship is very small, as
indicated by the predicted probabilities. Finally, poor renters do not appear to
be especially susceptible to displacement or elevated rates of mobility. Taken
together, the results would not seem to imply that displacement is the primary
mechanism through which gentrifying neighborhoods undergo socioeco
nomic change. Nevertheless, it is true that gentrification was related to dis
placement in this analysis, contrary to the findings of Vigdor (2002) and
Freeman and Braconi (2004).
The other control variables in the models of residential mobility perform
according to theoretical predictions for the most part. Age is an important
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
predictor of whether someone moves, as are family size, being a renter, gen
der, and permanent income. Besides gentrification, other geographic factors
were important determinants of mobility, with the vacancy rate associated
with higher rates of mobility and residents outside the Northeast being more
The model using displacement as a dependent variable closely parallels
the residential mobility model. The key difference is that length of residence,
marital status, and weeks unemployed last year did matter, whereas gender
and other race did not matter.
Thus far, the link between gentrification and displacement appears to be
modest. The evidence does not consistently show that higher rates of mobil
ity and displacement rates, although positively linked to gentrification, are
only modestly greater in gentrifying neighborhoods. Perhaps when a more
direct measure of the causal mechanism behind displacement is imple
mented, the smoking gun will surface.
Table 6 shows the relationship between rental inflation and residential
mobility. For the sake of brevity, only the independent variables are pre
sented. As was the case with the models using gentrification as independent
variables, the interaction terms that tested whether poor renters were espe
cially susceptible to displacement proved to be insignificant and hence were
dropped from the final models, but they are illustrated in Table A1 in the
The higher odds ratio suggests that the effects of rental inflation are con
sistent with the notion that escalating rents increase displacement. For both
Relationship Between Rental Inflation and Displacement and
Probability of Moving Probability of Being Displaced
Rental inflation 1.06* 1.40***
Predicted Probabilities
Lowest rental inflation third 0.14 0.013
Medium rental inflation 0.14 0.015
High rental inflation 0.15** 0.018**
*p < .01. **p < .05. ***p < .001.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
mobility and displacement, the odds ratios are significant and greater than 1.
Substantively, however, the incremental increases in displacement rates
associated with rental inflation are rather small. Consider the probability
of being displaced for someone residing in a neighborhood with a rate of
rental inflation of 8%, which corresponds to the 33rd percentile of rental
inflation—0.013%. This increases to 0.018% for someone residing in a
neighborhood that experienced rental inflation at the 99th percentile.
Taken together, the results suggest that although rental inflation is related
to displacement in gentrifying neighborhoods, the magnitude of the relation
ship is rather modest.
Because of the lack of a consensual operationalization of gentrification
and the admitted arbitrariness of the definitions used here, additional specifi
cations of gentrification were attempted to discern how sensitive the results
were to these alternative specifications. For the sake of brevity, only the gen-
eral patterns are discussed here, but the full results are available from the
author upon request. As was mentioned earlier, interaction terms testing
whether poor renters were especially susceptible to displacement proved
unfruitful. Because rental inflation did have a modest relationship to mobility
and displacement, rental inflation was substituted for housing price inflation
in the definition of gentrification described on pages 9 and 10. This alterna-
tive definition, however, resulted in substantively similar results as those
reported above. Another alternative specification attempted was to restrict
gentrifying neighborhoods only to those in which there was a relative
increase in housing prices, measured as an increase in a neighborhood
greater than or equal to the increase in housing prices in the surrounding met
ropolitan area. This approach, however, yielded substantively similar results.
The analyses reported above were also estimated separately for the 1980s and
1990s respectively. These decade-specific estimates were substantively simi
lar, except that evidence of gentrification being associated with lower mobil
ity was found in the 1990s.
To assess whether the relationship between
displacement/mobility and rental inflation is more aptly captured through a
nonlinear specification, rental inflation was specified in a nonlinear fashion
as a series of three and five dummy variables, respectively. Under these speci
fications each dummy variable represented a third quintile of the rental infla
tion distribution, respectively. These nonlinear specifications, however, did
not reveal a consistent statistically significant relationship between rental
inflation and displacement/mobility.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
The results presented here suggest that the relationship between gentrifi
cation and displacement is not especially robust. One strategy that soughtto
capture evidence of displacement through higher mobility rates did not re
veal a link to gentrification. Both displacement as an outcome and rental
inflation as an independent variable produced results that were more consis
tent with gentrification being linked to displacement. But here, the sizesof
the relationships were very modest. It should also be remembered that dis
placement is a subset of all moves and that mobility was not found to be
higher in gentrifying neighborhoods, implying that other types of moves
might actually be lower in these neighborhoods. Taken together, these empir
ical results provide little evidence that displacement is the engine of neigh
borhood change in gentrifying neighborhoods. Yet Tables 1 and 2 showed
that the neighborhoods defined here as gentrifying improving did indeed
improve in socioeconomic status throughout time. Certainly, anecdotal evi-
dence also describes gentrifying neighborhoods as becoming more upscale
throughout time. What explains this apparent paradox?
Although gentrification does not appear to be associated with increased
mobility or high levels of displacement, there are still other mechanisms
through which neighborhoods can change their character and become more
upscale. First, let us revisit the issue of mobility. The analyses presented
above speak to the question of whether households move. They do not
address the issue of where the households move. This could conceivably be
an important component of whether a neighborhood changes. For example,
a neighborhood where mobility is high but most of the movers remain in
the neighborhood would appear to change less than a neighborhood with less
mobility but where all of the movers move out of the neighborhood—
especially if the in-movers are noticeably different in the latter case. To the
extent that gentrification is associated with current residents no longer being
able to afford housing in their current neighborhood, one would expect to
find fewer intraneighborhood moves. Alternatively, to the extent that gentri
fication increases satisfaction with a neighborhood, current residents might
be more likely to want to stay in their present neighborhood and hence less
likely to move out of the neighborhood. We therefore have two competing
forces: gentrification acting to decrease intraneighborhood mobility because
of rising housing prices, and gentrification acting to increase intraneighbor
hood mobility because of increased residential satisfaction.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
To gauge the impact of these forces, the rate of intraneighborhood mobil-
ity, measured as the percentage of all moves in which the destination is the
same census tract, was compared across gentrifying, nongentrifying, and all
other urban neighborhoods. Intraneighborhood mobility was estimated for
all movers, those who did not attend college, and the poor, respectively. The
estimates illustrated in Table 7 show that the rate of intraneighborhood mo-
bility was typically lower in gentrifying neighborhoods. Across all three
groups, moves originating in gentrifying neighborhoods were more likely to
end outside of the neighborhood when compared to the counterfactual non
gentrifying neighborhoods. As Table 7 shows, the amount of intraneighbor
hood mobility is not inconsequential, ranging between 25% and 45% of all
moves. Altering this flow could have a substantial impact on the composition
of a neighborhood. This result suggests that gentrification may inhibit intra
neighborhood mobility and contribute to demographic change in that way.
Much discussion on neighborhood change, whether regarding gentrifica
tion or racial change, has focused on people moving out, that is, displacement
or White flight. But, as others have pointed out, neighborhoods are dynamic
entities, and who moves in can be just as important as who moves out in deter
mining neighborhood change (Gould Ellen 2000; Galster 1998). That raises
TABLE 7: Intraneighborhood Mobility by Neighborhood Type
Proportion Moving Outside of Neighborhood
Type of Neighborhood All (%) Poor (%) No College Attendance (%)
Gentrifying neighborhoods 63.6 71.2 64.4
Nongentrifying neighborhoods 57.4* 67.1* 55.7*
All other neighborhoods 63.3 75.9 62.9
40th Percentile
Type of Neighborhood All (%) Poor (%) No College Attendance (%)
Gentrifying neighborhoods 74.7 77.5 75.6
Nongentrifying neighborhoods 57.1* 67.3* 55.4*
All other neighborhoods 70.3* 75.8 75.7*
*p < .01.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
the question of whether a person’s socioeconomic status is related to the like
lihood of moving into a gentrifying neighborhood. A major concern regard
ing gentrifying neighborhoods is that these become neighborhoods that are
no longer available to low-income households. It appears that disadvantaged
households do not necessarily leave more quickly. But what of the option to
move in? We turn our attention this question in this section.
To determine how gentrification affects who moves into a neighborhood,
a series of bivariate multinomial regression models were estimated with
educational attainment, income, race, and poverty status as the indepen
dent variables. These multinomial regression models have the type of
neighborhood—gentrifying, nongentrifying, and all other metropolitan
neighborhoods—as the dependent variable. These bivariate analyses com
pare the likelihood of moving into a gentrifying as opposed to a nongentri
fying neighborhood along the dimensions described above. The key question
here is whether socioeconomic status is associated with moving into gen
trifying neighborhoods. Because we simply want to know how gentrification
is related to the characteristics of in-movers, there is no need to control for
other variables. To the extent that socioeconomic characteristics influence
who moves into gentrifying neighborhoods, the independent variables in this
series of regression models should be statistically significant.
Table 8 suggests that socioeconomic status does indeed influence who
moves into gentrifying neighborhoods. The third row illustrates the relation-
ship between various indicators of socioeconomic status and the likelihood
of moving into a median-defined gentrifying neighborhood as opposed to a
nongentrifying neighborhood. Odds ratios greater than 1 mean that someone
with that characteristic is more likely to move into a gentrifying neighbor
hood, whereas an odds ratio of less than 1 means someone is less likely to
move into a gentrifying neighborhood. Higher incomes and being White are
associated with an increased likelihood of moving into such a neighborhood,
whereas being Black is associated with a decreased likelihood. The pattern
evinced in the sixth row is even stronger. Here, being a college graduate,
being White, and having a higher income are associated with an increased
likelihood of moving into a 40th-percentile-defined gentrifying neighbor
hood. Conversely, being poor, without any college education, and Black
are negatively associated with moving into a median-defined gentrifying
These bivariate regressions show that in-movers to gentrifying neigh
borhoods are of higher socioeconomic status and more likely to be White,
characteristics normally associated with the gentry. The dynamics of a neigh
borhood change in gentrifying neighborhoods should thus be viewed as a
two-sided phenomenon. On one side is who moves out of the neighborhood,
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
TABLE 8: Relationship Between Socioeconomic Characteristics and Moving into Neighborhoods
Likelihood of Moving into Neighborhood
Compared to Nongentrifying Neighborhood Poor College Graduate No College Income White Black
Gentrifying neighborhood
0.83 1.18 1.05 1.01*** 1.93*** 0.41***
Other neighborhood
0.48*** 1.93*** 0.65*** 1.01*** 4.27*** 0.19***
40th Percentile
Gentrifying neighborhood
0.71*** 1.45*** 0.65** 1.01*** 2.26*** 0.37***
Other neighborhood
0.38*** 2.03*** 0.93 1.01*** 5.14*** 0.16***
a. The reference category here is nongentrifying neighborhoods. The odds ratios provide the likelihood of moving into the type of neighborhood indicated in
column 1 compared to moving into a nongentrifying neighborhood.
**p < .05. ***p < .001.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
the side that has garnered much of the attention in the literature on gentrifica
tion and displacement. The results presented here suggest that mobility out of
gentrifying neighborhoods is not necessarily dramatically different from
mobility out of other neighborhoods. Certainly, the results are inconsistent
with the notion that high rates of displacement always accompany neighbor
hood gentrification. These results also echo those of Freeman and Braconi
(2001), who found little evidence of displacement in gentrifying neighbor
hoods but did find in-movers to gentrifying neighborhoods to be of higher
socioeconomic status than current residents. On the other side of the gentrifi
cation process are the in-movers. The so-called gentry have attracted atten
tion in terms of describing who these people are. Overlooked perhaps is the
extent to which changes in the characteristics of in-movers could be the more
important force in determining the way that neighborhoods change.
That in-movers rather than out-movers are the driving force behind neigh
borhood change in gentrifying neighborhoods makes intuitive sense. People
are likely to be more sensitive to neighborhood characteristics when choos-
ing what neighborhood to move into rather than whether they should move at
all. This is because moving is costly in terms of time, money, and the possible
disruption of social ties and daily routines. Once people have made the deci-
sion to move, however, these costs take less prominence in the equation. The
characteristics of the destination neighborhood are then likely to be relatively
more important.
Gentrification remains a hot-button topic sure to set off debates and con
troversy about how it affects neighborhoods and the people residing there.
Certainly, many people recognize the possible benefits of gentrification:
increased amenities, improved public services, and rehabilitated housing. As
noted earlier, the fear of displacement has in the minds of many, however,
come to dominate all other concerns regarding gentrification. Here, I con
sider the implications of these findings from both a theoretical perspective
and a more pragmatic policy oriented perspective.
The results presented here indicate that the process of neighborhood
change associated with gentrification and revitalization more broadly isre
lated to displacement, albeit modestly. Who moves into the neighborhood
appears to be more important in explaining neighborhood change in gentrify
ing neighborhoods. While this analysis did not find lower mobility rates in
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
gentrifying neighborhoods as Freeman and Braconi (2004) and Vigdor
(2002) did, the results were consistent with earlier studies in illustrating that
neighborhoods can gentrify without widespread displacement. To be sure,
there are instances when the displacement of preexisting residents might
aptly describe the dynamics of change in a gentrifying neighborhood. But
when this process is viewed more broadly, it seems that the more typical
engine of neighborhood change is the altering of the characteristics of in-
movers and the lower rates of intraneighborhood mobility in gentrifying
neighborhoods. Other types of neighborhood change, particularly the notori
ous White flight associated with White-to-Black transitions, might be more
rapid and characterized by the rapid out-migration of preexisting residents
once some tipping point is breached (Crowder 2000). Gentrification, how
ever, is perhaps a more gradual process that, although displacing some,
leaves its imprint mainly by changing who moves into a neighborhood. For
students of neighborhood change, this is an important lesson to understand.
From a policy perspective, the implications are perhaps subtler. Gentrifi-
cation brings with it increased investment and middle-class households to
formerly forlorn neighborhoods. This could potentially enhance the tax base
of many central cities and perhaps increase socioeconomic integration as
well. After decades of disinvestment and middle-class flight, these benefits
from the gentrification should not be overlooked. The chief drawback has
been the inflation of housing prices in gentrifying neighborhoods. The
results presented here might tempt one to conclude that the lack of wide-
spread displacement means that concerns about the disappearance of afford-
able housing are overblown. But the fact that lower socioeconomic status
households are no longer moving into these neighborhoods implies a dimin
ishing of housing opportunities for some. Households that would have for
merly been able to find housing in gentrifying neighborhoods must now
search elsewhere. Whether suitable conditions are available elsewhere will
depend on the conditions of the particular housing market. But to the extent
that there is a shortage of affordable housing, it would seem to matter little if
those being affected are households who have to move because prices are
increasing or households find some options closed off because prices are
Moreover, although displacement may be relatively rare in gentrifying
neighborhoods, it is perhaps such a traumatic experience to nonetheless
engender widespread concern (Fullilove 2004). Consequently, the results
presented here still speak to the need for planners and policy makers to antici
pate the impacts that gentrification can have on housing affordability andto
plan accordingly.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
TABLE A1: Interaction Between Poor Renter and Gentrification
Moved Displaced
Median 40th Percentile Median 40th Percentile
Gentrifying neighborhood 0.95 0.83 1.36* 1.69***
Other neighborhood 0.89 0.92 1.81 2.05**
Gentrifying neighborhood *
poor renters 1.01 0. 73 1.35 1.28
Other neighborhood * poor renter 0.58* 0.585* 1.83 2.64*
Poor renter 1.58 1.63* 0.98 0.68
NOTE: Boldface means the item is statistically significant.
*p < .01. **p < .05. ***p < .001.
Interaction Between Poor Renters and Rental Inflation
Probability of Moving Probability of Being Displaced
Rental inflation 1.12 1.48***
Rental inflation*poor renter 0.55*** 0.93
Poor renter 1.47*** 2.16*
Medium rental inflation 1.24** 1.29
High rental inflation 1.11 0.79
Medium rental inflation*Poor renter 1.55* 0.46
High rental inflation*Poor renter 0.85 0.53
Poor renter 1.06 1.79*
NOTE: Boldface means the item is statistically significant.
*p < .01. **p < .05. ***p < .001.
1. Much thanks to Kurt G. Usowski of HUD for supplying these data.
2. There is evidence to suggest that substantial misreporting occurs when respondents of sur
veys like the PSID are asked about residing in subsidized housing (Shroder 2002). This misre
porting appears to be most problematic when respondents are asked to identify the type of hous
ing subsidy they receive rather than if they are receiving any housing subsidy at all. Because the
focus here is not on a specific type of housing subsidy, instead examining whether they receive
any subsidy at all, this misreporting error should not be fatal.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
3. The reader should note that tracts are classified based on their characteristics at the begin
ning of each decade. Thus, tracts designated as gentrifying in the 1980-1990 period are necessar
ily considered gentrifying in the subsequent decade.
Baldassare, M. 1982. Evidence for neighborhood revitalization: Manhattan in the 1970s. Journal
of Urban Affairs 4:25-37.
Beauregard, R. 1986. The chaos and complexity of gentrification. In Gentrification of the city,
edited by Neil Smith and Paul Williams, 35-55. Boston: Allen and Unwin.
Clay, P. L. 1979. Neighborhood renewal. Lexington, MA: Lexington Books.
Crowder, K. 2000. The racial context of White mobility: An individual level assessment of the
White flight hypothesis. Social Science Research 29 (2): 223-57.
Fielding, Elaine. 1994. Understanding why household move: A comprehensive analysis of rea
sons for moving. Research Report 94-305. Ann Arbor: University of Michigan Population
Studies Center.
Freeman, L., and Frank Braconi. 2004. Gentrification and displacement: New York City in the
1990s. Journal of the American Planning Association 70 (1): 39-52.
Fullilove, Mindy. 2004. Root shock. New York: Ballantine.
Galster, George. 1998. A stock/flow model of defining racially integrated neighborhoods. Jour-
nal of Urban Affairs 20:43-52.
Gould Ellen, Ingrid. 2000. Sharing America’s neighborhoods. Cambridge, MA: Harvard Univ.
Grier, G., and E. E. Grier. 1978. Urban displacement: A reconnaissance. Washington, DC: U.S.
Department of Housing and Urban Development.
Hammel, D. J., and E. K. Wyly. 1996. A model for identifying gentrified areas with census data.
Urban Geography 17 (3): 248-68.
Hamnett, C. 1991. The blind men and the elephant: The explanation of gentrification. Transac
tions, Institute of British Geographers 16:259-79.
Hartman, C. 1979. Comment on neighborhood revitalization and displacement: A review of the
evidence. Journal of the American Planning Association 45 (4): 488-92.
Henig, J. R. 1980. Gentrification within cities: A comparative analysis. Social Science Quarterly
61 (3): 639-52.
Hill, M. S. 1992. The panel study of income dynamics: A user’s guide. Thousand Oaks, CA: Sage.
Jaccard, J. S. 2001. Interaction effects in logistic regression. Newbury Park, CA: Sage.
James, Franklin J. 1977. Back to the city: An appraisal of housing reinvestment and population
change in America. Washington, DC: Urban Institute.
Lee, B. A., and D. C. Hodge. 1984. Spatial differentials in residential displacement. Urban Stud
ies 21:219-31.
Ley, David. 1980. Liberal ideology and the post industrial city. Annals of the Association of
American Geographers 70:238-58.
Lipton, G. 1977. Evidence of central city revival. Journal of the American Institute of Planners
(April): 136-47.
National Urban Coalition. 1978. Displacement: City neighborhoods in transition. Washington,
DC: National Urban Coalition.
Newman, Sandra J., and Michael S. Owen. 1982. Residential displacement: Extent, nature, and
effects. Journal of Social Issues 38 (3): 135-48.
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
Rose, D. 1984. Rethinking gentrification: Beyond the uneven development of Marxist urban the
ory. Environment and Planning D 2:47-74.
Rossi, P. H. 1980. Why families move. New York: Free Press.
Schill, M. H., and N. Richard. 1983. Revitalizing America’s cities: Neighborhood reinvestment
and displacement. Albany: State Univ. of New York Press.
Shroder, M. 2002. Does housing assistance perversely affect self-sufficiency? A review essay.
Journal of Housing Economics 11:381-417.
Smith, N. 1979. Toward a theory of gentrification: A back to the city movement by capital not
people. Journal of the American Planning Association 45:538-48.
———. 1996. The new urban frontier: Gentrification and the revanchist city. London: Routledge.
———. 1998. Gentrification. In The encyclopedia of housing, edited by W. Van Vliet--, 198-99.
London: Taylor & Francis.
Spain, D., J. Reid, and L. Long. 1980. Housing successions among Blacks and Whites in cities
and suburbs. Current Population Reports, Series P-23, no. 101. Washington, DC: U.S. Cen
sus Bureau.
Speare, A. 1974. Residential satisfaction as an intervening variable in residential mobility.
Demography 11 (2): 173-88.
Sumka, H. 1979. Neighborhood revitalization and displacement: A review of the evidence. Jour
nal of the American Planning Association 45:480-87.
U.S. Census Bureau. 2003. Educational Attainment in the U.S.. March 31. http://landview.
U.S. Department of Housing and Urban Development. 1979. Whither or whether urban distress.
Working paper, Office of Community Planning and Development. Washington, DC: U.S.
Department of Housing and Urban Development.
Vigdor, J. 2002. Does gentrification harm the poor? Brookings-Wharton Papers on Urban
Affairs: 133-73.
Lance Freeman is an assistant professor in the Urban Planning Program at Columbia
University. His research interests include neighborhood change, housing policy, and
racial/ethnic inequality in the urban United States. He has most recently published arti
cles on minority homeownership and the impacts of assisted housing on surrounding
at COLUMBIA UNIV on July 22, 2013uar.sagepub.comDownloaded from
... Some consider gentrification an inevitable consequence of the redevelopment process, while others find it less problematic (Atkinson, 2003). For example, some suggest that displacement cannot occur in new-build gentrification as there are no current residents to displace (Payne & Greiner, 2019) or that lower-income residents are not being displaced due to gentrification (Freeman, 2005;Sumka, 1980). Others contend that gentrification is necessary as it propels redevelopment in reaction to insufficient funding and subsidies to ensure equitable redevelopment (Williams, 1986). ...
... In this phase, data were collected for each city at the census tract level and the city level, including total population, population density, percentage of female population, percentage of people over 25 years with a bachelor's degree and above, percentage of white population, percentage Black population, median housing income, median house value, median gross rent as well as poverty level. There is disagreement on whether race is an integral part of gentrification (see Bates, 2013;Goetz, 2011) or independent of the process (Bostic & Martin, 2003;Freeman, 2005). This research uses race as gentrification often plays out along racial lines (Rucks-Ahidiana, 2021), historical and current patterns of disinvestment, structural racism, and redlining. ...
..., an online magazine, created a gentrification index by determining whether tracts eligible to gentrify (using variables that demonstrate lower economic values and/or vulnerable populations) gentrified (Maciag, 2015b). Using Freeman's (2005) widely used methodology, gentrifiable tracts are central city tracts populated by low-income households and have suffered disinvestment over time. This research adapted Governing's methodology to be used in mid-sized cities (see Appendix E for the complete Governing Methodology). ...
This study explores the policy tools mid-sized American cities use in their redevelopment efforts and determines the impact of these policy tools on the cities’ levels of gentrification. Theories of neighborhood change suggest factors that lead to decline, requiring redevelopment. Gentrification and revitalization are described as opposite ends of a redevelopment spectrum, where alleviating the harms of gentrification promotes revitalization outcomes. Policy tool types (protection, access, supply, and empowerment (PASE) are suggested to address gentrification’s harms. This study uses an exploratory sequential mixed methods research design. Phase 1 employs qualitative methods to create a typology of cities by analyzing pro-social redevelopment policies from a purposive sample of 103 mid-sized American cities. The typology is used as an independent variable to determine any relationship between policy tool use and gentrification. Study Phase 2 measures gentrification quantitatively, creating a gentrification index, and then tests for any statistical significance between the index and the city type. Data were collected from ACS 5-year estimates and the census for four decades, 1990, 2000, 2010, and 2019, to show changes in socioeconomic factors associated with gentrification. Using a pre-developed methodology, census tracts were assigned values to demonstrate changes in tracts over the four decades, creating a gentrification index. GIS analysis displayed low levels of gentrification; however, it revealed that cities were experiencing high rates of neighborhood change (decline and/or growth) instead of stability. ANOVA single-factor analyses failed to reject the null hypotheses; thus, there is no support that policy type impacts gentrification levels of cities. While there were no significant relationships between gentrification indexes and city type, the findings suggest more research is needed. Future research can add to the understanding of mid-sized American cities by examining the reasons for high decline or growth, the lack of adoption of pro-social policies, and challenge traditional methods of measuring gentrification.
... However, a qualitative study of antebellum Manhattan highlighted how parks have historically commanded higher real-estate premiums and been used to "clean up" disinvested areas (McNeur, 2017), suggesting this last point does not confirm green gentrification to be only a recent phenomenon. Quantitative gentrification studies have typically used indicators based on ethnicity, education, income, professional status, and housing costs to identify the inmoving of professionally employed, high-income, White individuals (Freeman, 2005;Ding et al., 2016). As mentioned above, much gentrification research is rooted in the US, resulting in the repeated use of certain variables. ...
... Beyond standard indicators, some have justified the inclusion of age (Romero and Harris, 2019), gender (Pearsall and Eller, 2020), and country of origin (Anguelovski et al., 2017). As in earlier quantitative gentrification studies (Ding et al., 2016;Freeman, 2005), some green-gentrification studies used criteria to determine a priori which spatial units are "gentrifiable" (Anguelovski et al., 2017;Rigolon and Németh, 2020;Shokry et al., 2020). Some studies used only income for this a priori designation, which could miss other factors that make an area susceptible to gentrification-such as building stock, historical status, proximity to other gentrifying areas/CBD. ...
Full-text available
This systematic literature review identifies and critiques methodological trends in green gentrification research (focusing on studies of vegetative greening) and provides suggestions for advancing this field. Findings reveal (1) research has largely focused on U.S. case studies; (2) early work employed qualitative methods but quantitative analyses have become more common; (3) little attention has been paid to the influence of greening characteristics/functions and non-greening factors on gentrification; (4) the mechanisms through which greening leads to gentrification are not well understood, particularly on the demand side; and (5) despite being the main concern of green gentrification, displacement has not been well-documented.
... Er zeigt, dass 38% der Arbeiterklassehaushalte zwischen 1981 und 1991 aus Gentrifizierungsvierteln weggezogen sind (für weitere jüngere Studien, die Verdrängungsraten quantifizieren, siehe auch Freeman und Braconi 2004sowie Newman und Wyly 2006. Die baulich-immobilienwirtschaftliche Seite wird in den meisten dieser Studien nicht operationalisiert und in die Analyse einbezogen (für eine Ausnahme siehe Freeman 2005). ...
Full-text available
Vor dem Hintergrund der steigenden Relevanz und Aufmerksamkeit für Gentrifizierung wurden in den letzten Jahren vermehrt Methoden entwickelt, um Gentrifizierungsgebiete zu identifizieren. Die vorliegende Studie entwickelt, aufbauend auf dem internationalen Forschungsstand, erstmals eine Methode zur quantitativen Identifizierung solcher Gebiete in Wien und wendet diese auf Basis von kleinräumigen sozioökonomischen und immobilienwirtschaftlichen Daten für den Zeitraum 2011 bis 2016 an. Neben den empirischen Befunden bietet die Studie damit eine methodische Innovation gegenüber der vorhandenen Forschung zu Gentrifizierung in Wien, die sich bis dato auf Fallstudien einzelner Viertel oder Sektoren am Wohnungsmarkt beschränkt. Darüber hinaus leistet die Studie einen Beitrag zur breiteren sozialgeographischen Stadtforschung in Wien, die bisher vorwiegend auf sozioökonomische Veränderungen fokussiert und immobilienwirtschaftliche Veränderungen nicht näher in den Blick nimmt. Die Ergebnisse werden vor dem Hintergrund der vorhandenen Literatur diskutiert und offene Fragen für die zukünftige Forschung zu Gentrifizierung in Wien identifiziert.
This study examines links between gentrification and neighborhood health. Gentrification is associated with decreases in neighborhood poverty and crime, increases in amenities and services, among other benefits—all identified as structural determinants of health. However, gentrification is also associated with population-level replacement of the existing community, or threats thereof, thus raising the possibility that community improvements via gentrification may negatively impact current residents. Combining census data from the ten largest MSAs in the U.S. with tract-level estimates from the CDC-PLACES Project from 2013-14 to 2017-18, we perform spatial regression models that explore how the changing socioeconomic conditions in gentrifying neighborhoods correlate with changes in neighborhood health, both within those neighborhoods and in other areas that may be impacted. We find significant differences between gentrifying and non-gentrifying neighborhoods in their associations with neighborhood health. The sociodemographic changes occurring in gentrifying neighborhoods tend to correspond with simultaneous decreases in aggregate health risk behaviors and negative health outcomes. However, these neighborhood changes are heterogeneous and complex. Whether and how neighborhood health changes alongside other components of neighborhood change depends on the initial racial composition of the neighborhood—whether gentrification occurs in majority Black, Hispanic, or White neighborhoods. Our findings thus provide preliminary evidence that the changes accompanying gentrification do extend to neighborhood health, but the direction of influence varies by neighborhood composition, type of sociodemographic change, specific health outcome, and spatial spillover. We discuss theoretical implications for future work addressing the mechanisms driving changes in neighborhood health, and potential approaches that differentiate policy responses.
There has been a renewed interest in moving back into cities, which are close to business districts and offer affordable and unique housing. However, the available housing is limited and may require renovation as available housing close to the citycenter is often located in more economically disadvantaged neighborhoods. Property repairs and upgrading contributes not only to the improvement of individual owners’ properties, but the neighborhood overall. This influx of new residents and subsequent investment in housing can impact neighborhood crime, but the majority of the research on housing and crime has focused largely on home mortgage loans. The current study extends the housing investment literature by using an underutilized data source: home improvement and refinance loans, which signal physical improvement in the housing stock. This process may be different for neighborhoods that have a higher prevalence of minority residents as historically these residents have been subjected to inequitable lending practices. The current study examines how revitalization efforts in Cleveland, Ohio have influenced crime rates for the years 2007 through 2017 with special attention paid to the interplay of neighborhood racial composition and home improvement loans. Results from fixed-effects panel analysis reveal that home improvement loans are associated with an increase in property but not violent crime rates overall. Splitting neighborhoods into predominantly Black versus all other neighborhoods, however, results show that higher rates of home improvement loans are associated with lower violent crime rates, but the effect is tempered in Black neighborhoods. This suggests that the relationship between home improvement loans and violent crime is more complicated and varies by neighborhood composition.
Gentrification is happening in cities all across the United States. Consequently, some Black communities that were intentionally segregated and under-resourced are experiencing capital investments and demographic changes. These gentrification-induced racial and socioeconomic shifts impact many local institutions, namely school districts. Given this, there is an emerging body of research on schools and gentrification. However, less research has examined the actions of school districts as institutional actors in gentrification. This study examines how two school districts’ actions mediate school gentrification. Using a theorization of gentrification as a process of racial capitalism, we draw on interviews with 26 principals across both districts. Our findings suggest that districts’ actions influence school gentrification by mediating the movement of Black and other youth of color to various schools through cycles of differential investments across the districts. We conclude with implications for future research.
The issue of neighborhood revitalization and displacement certainly does have a déjà vu quality to it. While government officials, academic housing analysts, and even neighborhood groups seem to approach the problem as if it were a brand new phenomenon, in reality the recent history of displacement under urban renewal, the interstate highway system, and other government programs is very relevant. What is remarkable in the spate of literature that recently has emerged on the issue—of which Sumka's article is quite representative—is the failure to acknowledge that history and the lessons it might offer.
This paper critically reviews the major theories of gentrification which have emerged over the last 10 years and the debate which has surrounded them. It argues that the reason why the gentrification debate has attracted so much interest, and has been so hard fought, is that it is one of key theoretical battlegrounds of contemporary human geography which highlights the arguments between structure and agency, production and consumption, capital and culture, and supply and demand. It also argues that each of the two major explanations which have been advanced to account for gentrification (the rent gap and the production of gentrifiers) are partial explanations, each of which is necessary but not sufficient. Finally, it argues that an integrated explanation for gentrification must involve both explanation of the production of devalued areas and housing and the production of gentrifiers and their specific consumption and reproduction patterns.
Empirical research on gentrification suffers from a dichotomy between richly detailed neighborhood case studies and macro-scale, census-based analyses, perpetuating uncertainty over the extent and timing of gentrified areas in American cities. We develop a model relating tract-level census statistics to the results of a detailed field survey of 24 census tracts in Minneapolis-St. Paul. We use stepwise and canonical discriminant analysis to select nine variables distinguishing gentrified neighborhoods and to classify all central-city tracts for each decade between 1960 and 1990. Results indicate a moderate level of overall accuracy, and the model is more than 90% accurate in distinguishing areas of heavy reinvestment from stable, middle-class districts. Compared with other techniques, our approach more accurately distinguishes gentrification from other types of inner-city redevelopment, providing a useful tool for identifying the phenomenon with a measurable degree of precision.
A new ideology of livability in urban development changed the Vancouver landscape between 1968 and 1978. The agents of liberal ideology were a new elite of professional, technical, and administrative workers whose consolidation coincided with Vancouver's transition toward a service oriented postindustrial city. This group founded an urban reform party which assumed political power in 1972. They challenged the commitment to growth, boosterism, and the city efficient held by former civic administrations, presenting in its place a program of apparently humane, socially progressive, and aesthetic urban development. Despite some significant successes, the new ideology was also elitist and has generated new problems of social justice, giving rise to a countervailing political movement in the late 1970s. Except in special circumstances it seems the ideology of the livable city is rarely compatible with criteria of social equity or economic efficiency.