ETHNIC ENCLAVES AND THE ECONOMIC SUCCESS
OF IMMIGRANTS—EVIDENCE FROM
A NATURAL EXPERIMENT*
P
ER
-A
NDERS
E
DIN
P
ETER
F
REDRIKSSON
O
LOF
Å
SLUND
Recent immigrants tend to locate in ethnic “enclaves” within metropolitan
areas. The economic consequence of living in such enclaves is still an unresolved
issue. We use data from an immigrant policy initiative in Sweden, when govern-
ment authorities distributed refugee immigrants across locales in a way that we
argue is exogenous. This policy initiative provides a unique natural experiment,
which allows us to estimate the causal effect on labor market outcomes of living
in enclaves. We nd substantive evidence of sorting across locations. When sorting
is taken into account, living in enclaves improves labor market outcomes for less
skilled immigrants: the earnings ga in associated with a standard deviation in-
crease in ethnic concentration is 13 percent. Furthermore, the quality of the
enclave seems to matter. Members of high-income ethnic groups gain more from
living in an enclave than members of low-income ethnic groups.
I. I
NTRODUCTION
In mos t countries, immigrants tend to be spatially c oncen-
trated [Stark 1991]. This is particularly true for the United
States, where almost three-quarters of immigrants resided in
only six states in 1998; see Borjas [1999]. It is also true for
Sweden, where the share of the foreign-born population living in
the three largest metropolitan areas outstripped the share of the
native population by 18 percentage points in 1997. In this paper
we are concerned with the labor market consequences of this
* We have beneted from th e useful comments of two anonymous referees,
Joshua Angrist, George Borjas, Marianne Bertrand, Kenneth Carling, Jan Ek-
berg, Edward Glaeser, Claudia Goldin, Peter Hedstro¨m, Lawrence Katz, Magnus
Lo¨fstro¨m, Regine Riphahn, Dan-Olof Rooth, seminar participants at European
University Institute, Go¨teborg University, IIES, SOFI, Stockhol m School of
Economics, Tilburg Uni versity, Tinbergen Institute, University College London,
Uppsala University, the CEPR conference on “Marginal Labour Markets in Met-
ropolitan Areas,” the 2000 meeting of EALE/SOLE, the NBER Summer Institute
2000, and the CEPR European Summer Symposium in Labour Economics 2000.
We thank Lisa Fredriksson for expert data assistance. We also thank Sven
Hjelmskog, Stig Kattilakoski, Christina Lindblom, Anders Nilsson, Kristina
Sterne, and Lena Axelsson of the Immigration Board and Anna Gralberg of the
Ministry of Culture, who generously found time to answer our questions. This
research has been partly nanced through a grant from the Swedish Council for
Work Life Research (RALF).
© 2003 by the President and Fellows of Harvard College and the Massachusetts Institute of
Technology.
The Quarterly Journal of Economics, February 2003
329
location pattern. In particular, we as k the question: how does
living in ethnic enclaves affect labor market outcomes?
There are several competing hypotheses about the effects of
living in enclaves. One hypothesis is that residing in an enclave is
associated with a lower rate of acquisition of host country skills,
e.g., language. Another hypothesis is that the enclave represents
a network that increase s the opportunities for gainful trade in the
labor market, e.g., by disseminating infor mation on job opportu-
nities. So, the causal effect of living in an enc lave is a priori
ambiguous in sign.
Estimating the causal effect of enclaves is no easy task.
Because of sorting, treating the ethnic compositio n of an area
within a city as exogenous is problematic. Residential location is
a household choice variable, and this choice is very likely inu-
enced by unobserved attribute s also affecting labor market out-
comes. Recent research has made progress in addressing this
issue. One approach is to use the variation across metropolitan
areas arguing that sorting is less problematic in thi s case; see
Bertrand, Luttmer, and Mullainathan [2000], Cu tler and Glaeser
[1997], Dustmann and Preston [1998], and Gabriel and Rosenthal
[1999]. Another approach is to use pa rental choices of neighbor-
hoods, where the assumption is that this choice is exogenous wit h
respec t to the outcome of the offspring; Borjas [1995] is an exam-
ple. In general, these studie s suggest that disadvantaged groups
such as blacks and immigrants ar e hurt by segregation.
Although the identication strategies briey reviewed above
are less susceptible to sorting problems, we think that the valid-
ity of the implicit exogeneity assumptions is an open question.
For instance, differences in unobserved individual attributes
across metropolitan areas still remain a concern.
In this paper we take a different approach. Our exogenous
sourc e of variation comes from a Swedish government policy
concerning the initial location of refugee immigrants.
1
This policy
was viable between 1985 and 1991. Government authorities
place d refugees in localities that were deemed suitable according
to certain criteria. In practice, the ava ilability of housing was the
all-important factor. Our maintained hypothesis is that, because
of the policy, the initial location of immigrants is independent of
1. The Katz, Kling, and Liebman [2001] analysis of Moving to Opportunity in
Boston is another example where a policy experiment is used to identify the causal
effect of interest.
330 QUARTERLY JOURNAL OF ECONOMICS
unobservable individual characteristics. Hence, this “natural ex-
periment” enables us to reexamine the question of the economic
consequences of living in enclaves.
The government settlement policy had real consequences for
immigrant location. This is illustrated in Figure I, where we plot
the share of the immigrant inow and the immigrant stock tha t
resides in Stockholm and the north of Sweden, res pectively. Prior
to 1985, refugees were allowed to settle in a neighborhood of their
own liking. In 1985 the immigrant shares in Stockholm and the
north of Sweden stood at 36 and 5 percent, respectively. By 1991
the share living in Stockholm had been reduced by more than 3
percentage points, while the share residing in the north increased
by 2 percentages. Thus, the policy initiative c learly increased the
dispersion of immigrants across Sw eden.
Ou r results can be briey summarized as follows. We nd
pervasive evidence of sorting across local labor markets. In cases
where the size of the enclave matters for outcomes, the estimates
that account for sorting differ drastically from those tha t do not.
For low skilled immigrants, estimates that suffer from sorting
bias associate an earnings loss of 4 percent with a standard
deviation increase in ethnic concentration. Estimates that do not
F
IGURE
I
Shar e of Non-OECD Immigrant Inow (solid) and Stock (da shed) Located in
Stockholm and in the North of Sweden, Respectively, 1978–1997
“Stockholm” refers to the county of Stockholm, and “North” to the six northern-
most counties of Sweden. Authors’ calculations using the LINDA immigrant
sample.
331ETHNIC ENCLAVES AND ECONOMIC SUCCESS
suffer from this problem suggest an earnings increase of almost
13 percent. Furthermore, the earnings gains are increasing in the
quality of the enclave. Members of high-income ethnic groups
gain more from living in an enclave than members of low-income
ethnic groups.
The remainder of the paper is outlined as follows. By way of
background, Section II gives a description of the institutional
setting and discusses whether the policy introduced in 1985 pro-
vides as a natural experiment. In Section III we outline a simple
framework that we use as a guide to speci cation and interpre-
tation. We then turn to the empirical analyses. Section IV de-
scribes the data. In Section V we report estimates of the effects of
ethnic concentration on earnings. Section VI concludes.
II. T
HE
I
NSTITUTIONAL
B
ACKGROUND
Sweden and the United States are sometimes portrayed as
being opposite extremes in terms of the scope of the welfare state
and the functioning of the labor market. Some aspects of the
immigration experience are, however, remarkably similar. The
share of the foreign born in the total population is about the same.
In 1997, 11 percent of the Swedish population was foreign born.
By comparison, the foreign born population in the United States
amounted to 10 percent in 1998. Moreover, the probability of
residing in an “ethn ic neighborhood” is of a similar magnitude in
the tw o countries.
2
Further, in both countries, immigrants from
developing countries are more likely to live in an ethnic neigh-
borhood. Of course, there are also differences between the two
countries. For in stance, immigrant source countries are obviously
different. Also, the share of refugees in the total inow of immi-
grants over the past 30 year s has been substantially larger in
Sweden.
Ou r analysis concerns refugee immigrants, and we now turn
to presentin g the Swedish refugee policy in more detail. U nfor-
tunately, there is very little documentation about the practical
implementation of the placement policy. Therefore, part of the
2. An ethnic neighborhood is dened as a neighborhood where the share of
the ethn ic group residing in the neighborhood is at least twice as large as the
share of the ethnic group in the population. According to Borjas [1999], 48 per-
cent of individuals with non-U. S. ancestry resided in an enclave in 1979. Doing
an analogous calculation for Sweden, we nd that 42 percent of the average
rst-generation immigrant lives in an ethnic neighborhood in 1997; see Edin,
Fredriksson , and Åslund [2001] for further details about this calculation.
332 QUARTERLY JOURNAL OF ECONOMICS
information is based on interviews with placement ofcers and
other of cials of the Immigration Board.
II.A. The Swedish Refugee Placemen t Policy
3
The placement policy was introduced in 1985. The Immigra-
tion Board was then given the responsibility of assigning refugee
immigrants to an init ial municipality of residence. The Boar d was
to place all political immigrants, except those who arrived for
family reunication reasons.
The introduction of the placement policy was a reaction to
immigrant concentration in large cities. The idea was to distrib-
ute asylum seekers over a larger number of municipalities that
had suitable characteristics for reception, such as educational
and labor market opportunities. At rst, the intention was to sign
contracts with about 60 municipalities, but due to the increasing
number of asylum seekers in the late 1980s, a larger number
became involved; in 1989, 277 out of Sweden’s 284 municipalities
participated. The factors that initially were supposed to govern
placement were more or less abandoned. Instead, the avail ability
of housing became the deciding factor.
Formally, the policy of assigning refugees to municipalities
was in place from 1985 to 1994. The strictest appl ication of the
assignment policy was between 1987 and 1991. F or our purposes,
this is the most attractive time period, since it contained very
small degree s of freedom for the individual immigrant to choose
the initial place of residence. During 1987–1991 the placement
rate, i.e., the fraction of refugee immigrants assigned an initial
municipality of residence by the Immigration Board, was close to
90 percent.
II.B. From Port of Entry to Placement
As a way of describing the placement policy during 1987–
1991, we describe the handling of a typical asy lum seeker from
the border to the nal placement. An asylum seeker was placed in
a refugee center while waiting for a decision from the immigra-
tion authorities. Refugee centers were distributed all over Swe-
den, and there was no correlation between the port of entry and
the location of the center. However, immigrants were sorted by
native language when placed in centers.
3. This section draws on The Committee on Immigration Policy [1996] and
The Immigration Board [1997].
333ETHNIC ENCLAVES AND ECONOMIC SUCCESS