Policy Influences on Refugee Resettlement Patterns
Jacqueline Desbarats
Introduction
Over the last two decades, political refugees have accounted for an increasing share
of U.S. immigration. From a mere 12 percent in 1965-70, the proportion of total admissions
made up by refugees rose to 23 percent in 1975-80 (Teitelbaum 1980). A consequence
of this sudden surge of refugee admissions in the late 1970s has been the emergence
of geographic distribution issues as a salient concem of legislators, policy makers,
and others who implement resettlement programs. This concem found expression in the
increasing frequency with which questions about refugee settlement pattems were raised at
Congressional hearings on refugee legislation (Forbes 1985b:23). It was formalized in the
1980 Refugee Act's mandate that the anticipated social, economic, and demographic
impact of refugee admissions in the U.S. be considered during the consultative process
determining annual numerical ceilings for refugee admissions (U.S. Department of Health
and Human Services 1981:1). The Office of Refugee Resettlement (ORR), created by the
Act within the U.S. Department of Health and Human Services, was given responsibility
for assessing the refugees' impact on individual communities and for reporting annually to
Congress on the refugees' employment statistics, their use of assistance programs, and their
geographic distribution (U.S. Department of Health and Human Services 1981).
In the long run, refugee residency patterns are affected not only by initial placements
but also by post-resettlement moves. Thus, the early focus on monitoring patterns of primary
placement gradually expanded to secondary migration patterns. A 1982 examination
of the domestic resettlement program conducted by the Surveys and Investigations Staff of
the House Appropriation Committee explicitly asked questions about the incidence of
secondary migration. In the same year, the Amendments to the Refugee Act specifically
directed ORR to compile and maintain data on refugee secondary migrations within the
U.S. (U.S. Department of Health and Human Services 1983:11).
Beyond having obvious pragmatic significance, refugee settlement patterns raise
some theoretical issues as well. Virtually all recent groups of immigrants to the U.S. have
initially settled in clustered patterns before yielding to the centrifugal forces of geographic
dispersal. This spatial sequence has commonly been attributed to the changing roles of the
factors affecting the newcomers' well-being, such as proximity to relatives and access to
compatriots, prospects for employment and housing availability, familiar climate, and
desirable environment. To be sure, the locational choices of individual refugees, just like
those of other immigrants, are influenced by the pull of social networks, economic opportunities,
and ecological conditions, and the geographic distribution of refugee groups inevitably
reflects these relatively well-documented sociocultural, economic, and environmental
factors. But in addition, refugee resettlement is subject to such a degree of administrative
control that refugees initially face much greater locational constraints than do "voluntary"
immigrants, and are subject to a somewhat different mix of locational incentives. Because
the global forces shaping the geography of refugee groups are more complex than those
KAS Papers
shaping the geography of "voluntary" immigrants, refugee settlement patterns cannot be
adequately explained without also considering explicitly how they are affected by policy
interventions.
From both an applied and a theoretical standpoint, then, it is useful to understand
how policy measures affect the geographic distribution of refugee groups. This question is
addressed in the present paper, which outlines the impact of federal- and state-level resettlement
and assistance policies on the locational choices of individual refugees and on the
dispersal or concentration of refugee groups. The effects of selected policy decisions are
illustrated with data on Indochinese refugee migrations.
Refugee Placement Policies
Refugee primary placement and secondary migration take place within the context of
a legislative framework that structures the desirability of locational options to an extent
rarely seen in the case of "voluntary" immigrants. Public policies influence refugee settlement
patterns in two ways: through initial placement and through refugee assistance.
Placement policies have consistently aimed at the dispersal of refugees over the national
territory, but assistance policies, though ostensibly neutrl in a geographic sense, have
sometimes promoted inadvertent concentration. Unintended spatial outcomes have
occurred because the interplay among public policies formulated and implemented at various
geographic levels of institutional decision-making has been quite complex and its
consequences somewhat unpredictable. Federal legislation on refugee admission, resettlement,
and assistance ultimately sets the bounds for state and local policy implementation
relative to refugees, and this, in turn, may affect the relative desirability of altemative
refugee destinations at both the regional and the local scales.
The Federal Level
As is common with new immigrants, virtually all incoming refugees prefer to live
with relatives, within a community of compatriots. Most refugee-receiving communities,
on the other hand, prefer to avoid the negative effects of large concentrations of newcomers
(Desbarats and Holland 1983:26). In recognition of this dual need, federal policies
on initial placement have attempted to strike a satisfactory balance between these two
conflicting considerations, though with diminishing success over time.
Refugees do not usually choose their initial destinations in the United States.
Although they may express their locational preferences, they are actually placed with
preselected sponsors by the Voluntary Agencies (Volags). Initial resettlement patterns,
therefore, are essentially the result of federal policies on refugee admission and placement,
implemented by the Volags. For the last forty years, official U.S. policy on initial
placement has advocated the dispersal of refugees throughout the country, with the expectation
that dispersed placement would speed up adaptation and spread the refugees' impact
more evenly among states and local areas. This policy was followed with Hungarian and
Cuban refugees, and later, with Southeast Asian refugees. In order to implement the
dispersed placement policy when the first Indochinese influx began, the Volags had to
select initial resetdement locations that did not necessarily correspond to the preferences
expressed by the refugees. Thus, less than half of the Indochinese refugees resettled in
1975 were placed in the state of their choice (Baker and North 1985).
When the resettlement of the second wave of refugees began in 1979, the policy of
dispersed placement was still enforced, but its effectiveness was being gradually diluted by
the increasing role of family reunification in the admission and resettlement of new
refugees. While a mere fifteen percent of the Vietnamese resettled in 1975 had been
so Desbarats
Refugee Resettlement
sponsored by family members, this figure shot up to 95 percent for those resettled in 1979
(U.S. Department of Health and Human Services 1982:31). Because the majority of the
new arrivals joined family members already in the U.S., most of them were placed with
their closest kin. This process increased the degree of geographic concentration, especially
in areas that already had substantial numbers of Southeast Asian refugees. The proportion
of refugees initially resettled in California, for instance, went up from 19 percent in 1975,
to 24 percent in 1978, and to 31 percent in 1981 (Desbarats 1985:526-27).
Throughout 1982, the uneven geographic distribution of refugees-in particular the
large concentrations of refugees in certain localities-and its effect on the placement of
future arrivals remained a major concern of federal policy makers (U.S. Department of
Health and Human Services 1983:27). Resettlement in communities with substantial
refugee concentrations places a serious strain on the capacity of these impacted communities
to respond to the needs of new arrivals and is thought to interfere with the refugees'
ability to become self-sufficient. The existence of a causal link between geographic concentration
and the increasing rates of public assistance dependency among the Indochinese
refugee population has even been suggested (U.S. Department of Health and Human Services
1982:22; Yarling 1985:4). The need to avoid the formation of other impacted areas
and to redirect the refugees away from areas of high concentration therefore led ORR to
launch a deliberate effort to alter the geographic distribution of incoming refugees (U.S.
Department of Health and Human Services 1982:22). Two distinct though complementary
strategies were proposed to achieve this goal: dispersing "free cases," i.e., those refugees
whose resettlement location was not predetermined by the location of family members
already in the U.S.; and stemming secondary migration, or at least, deflecting it away from
impacted areas.
One of the most innovative efforts in the search for new placement strategies was
the Khmer Guided Placement Program designed in 1981 to place 8,500 Khmer refugees in
twelve separate sites. This program departed somewhat from the standard guidelines of
dispersed placement in that it identified a few communities nationwide that had volunteered
to be hosts to small clusters of Cambodian refugees. Areas with large concentrations
of refugees and strained resources were specifically excluded from the project (U.S.
Department of Health and Human Services 1982:22). The program did have some positive
effects, as it achieved a reduction in the proportion of Cambodians resettled in California,
from 35 percent in 1980 to 18 percent in 1981 (Forbes 1985a:13). But it also had some
unintended consequences. Thus, the city of Providence, Rhode Island, which was not a
program participant, ended up with a Cambodian community twice as large as the largest
cluster site (Yarling 1985:6).
In another attempt to place free cases in favorable resettlement sites, ORR developed
the Favorable Altemate Sites Pilot Demonstration of Arizona and North Carolina, two
states that had initially registered lower-than-average densities of primary placement (Desbarats
1985:527-528). Along the same lines, the more recent National Discretionary Projects
have been designed to prevent the resettlement of free cases in areas of high impact
and to prohibit welfare recipients from acting as sponsors for incoming refugees (U.S.
Department of Health and Human Services 1983:28).
Since July 1982, formal placement policies have not merely emphasized the importance
of improving the quality of initial refugee placement. They have also sought to
reduce the incentives that might lead refugees to move away from their initial resettlement
sites. Secondary migration is hardly a new phenomenon among refugees. The displaced
persons admitted to the U.S. in the 1940s and 1950s, for instance, subsequently moved
away from Northeastern and Southern states toward the Pacific Coast (International Sociological
Association 1955). More recently, Cuban refugees recongregated in the Miami
Nos. 65-66 51
KAS Papers
area just a few years after their initial inflow. In the same way, Indochinese refugees have
voted with their feet in substantial numbers, largely as a reaction to the dispersed placement
policy. Dissatisfied with the climate and/or the social makeup of the communities in
which they had been placed, many of the new arrivals started to move, in an effort to
reconstitute communities of compatriots in more compatible climates (Barnes 1977), and
the late 1970s was a period of intense secondary migration. ORR, while conceding that
secondary migration is not in itself a negative phenomenon, has nonetheless recognized
that secondary moves that occur immediately or soon after a refugee is initially resettled
can waste scarce resources at the site of initial placement. Such movement has a greater
potential still to create further adverse impact if the refugees move with other welfaredependent
refugees to the new site, or into a community whose resources are already overtaxed
(U.S. Department of Health and Human Services 1982:21-22).
The State Level
Although the initial placement of the refugees are to a large extent shaped by federal
policies, they are also affected by the states' willingness to play an active role in the resettlement
process. A number of states, at some time, have assumed the role of sponsor and
acted as resettlement agencies, in particular the states of Washington, New Mexico, Indiana,
Missouri, Maine, Oklahoma, Iowa, and Idaho. Especially noteworthy is the effort of
the state of Iowa, whose participation, begun in September 1975, has continued for close
to a decade through a contract with the U.S. Department of State. Between 1975 and
1982, the Iowa State Refugee Service Center resettled close to 4,000 Southeast Asian
refugees, a large proportion of whom were hill people of the Thai Dam tribe closely
related through kinship ties. Their resettlement in a relatively clustered form has been
praised for its effectiveness, an implicit demonstration that, in some cases, it may be desirable
to depart from the dispersed placement policy (U.S. Department of Health, Education,
and Welfare 1979:39-40). On a smaller scale, and over a shorter time period, the state of
Idaho also acted as a resettlement agency, resettling two hundred Indochinese in 1981.
Finally, some states have become poles of attraction through the special training programs
that they set up for refugees in the professions. For example, the initiatives of such
states as Oklahoma, California, Pennsylvania, Florida, Nebrska, Arkansas, and Texas for
refugee physicians and dentists have been significant in altering the regional pattern of
resettlement. In 1975, all seven states had placement rates above the national average, as
measured by the number of resettled refugees per 100,000 state residents (Desbarats
1985:529).
Overall, then, even though federal placement policies have generally aimed at geographic
dispersal, some exceptions have been made in the case of specific groups. In
1975, 26 percent of the entering refugees would have needed to be placed in a different
state for their distribution to coincide with that of the American population. But, by 1980,
this proportion had gone up to 33 percent (Desbarats 1985:527). The effectiveness of
attempts at dispersal thus registered a slight decline over that five-year period.
Refugee Assistance Policies
The majority of the refugees start off life in the U.S. on public assistance. The need
to meet the special transitional needs of the refugees without creating inequities vis-a-vis
other disadvantaged populations has therefore become a salient consideration in the formulation
of refugee assistance programs (Forbes 1985b:2). But these programs have assumed
a significant role in shaping the system of incentives and disincentives to which refugees
respond. Secondary migration, in particular, has proved quite responsive to programmatic
52 Desbarats
Refugee Resettlement
changes in assistance policies.
Federal Policies on Domestic Assistance
Because the federal government reimburses states and local areas for the costs
incurred in connection with refugee programs, federal policies have affected both the availability
of assistance benefits and the financial level of these benefits. With the gradual
downturn in the federal responsibility for the cost of refugee resettlement, the mandated
period of federal reimbursement of the refugee costs of states and local areas has been
shortened. In turn, the resulting increase in the fiscal responsibility of state and local
govemments has meant a greater degree of geographic differentiation in assistance availability.
Federal policy on domestic resettlement was initially set by the Indochinese Migration
and Refugee Assistance Act of 1975, an ad hoc legislative initiative that was to be
effective for two years. Cash assistance to Indochinese refugees was to be provided on the
same basis as for other state residents, i.e., based on the same income limitations and payment
levels as apply in a State program of Aid to Families with Dependent Children
(AFDC). Because of their special circumstances, however, refugees were granted a waiver
of the family composition requirements normally applied to AFDC program recipients, so
that single refugees, as well as those in intact families, could receive benefits which would
not normally be available under AFDC. Refugee eligibility for Refugee Medical Assistance
(RMA) was subject to the same financial requirements normally applied in a state's
Medicaid program. One hundred percent of the expenses incurred by the states for the
Indochina Refugee Assistance Program (IRAP) were reimbursed by the federal government.
As of 1979, the states were given a broader role in the resettlement program through
an expansion of the scope of the federally reimbursed services that they were allowed to
provide for refugees. The 1980 Refugee Act created ORR as the federal agency with
overall responsibility for domestic resettlement activities, to provide resettlement assistance
to the refugees through a State-administered program. The Act authorized ORR to reimburse
states for up to 100 percent of the cost of providing cash and medical assistance to
eligible refugees, through the institution of a state plan mechanism for the provision of
refugee assistance (U.S. Department of Health and Human Services 1981:1). Thus, out of
the $672 million appropriated to ORR in Fiscal Year 1981, $516 million was used for the
State-administered programs, for the cost of providing cash and medical assistance and
social services to eligible refugees (U.S. Department of Health and Human Services
1982:6). In order to receive cash assistance, a refugee individual or family had to meet
the income and resource eligibility standards of his state of residence. Needy refugees
could be eligible for AFDC or for Refugee Cash Assistance (RCA), with the rules for
RCA approximating those for AFDC, except that RCA was available regardless of family
composition (U.S. Department of Health and Human Services 1981:9).
The year 1981 signaled the first tnajor reduction in the federal involvement of stateadministered
resettlement programs. Effective April 1, 1981, federal reimbursement to the
states was subject to a time limitation of 36 months from the time a refugee had entered
the country. Refugees who did not become self-supportive within that time frame would
become the fiscal responsibility of state- or county-funded programs. The rationale behind
the 36-month limit was that it would increase the states' incentives to devise service systems
likely to lead to rapid economic self-sufficiency among the refugees. According to
estimates of the New Transcentury Foundation, this change affected approximately 40,000
refugees on RCA. Time-expired refugees who were eligible for AFDC or Supplemental
Nos. 65-66 53
KAS Papers
Security Income (SSI) would still receive benefits by applying for existing AFDC programs
in their state of residence if they met the same eligibility standards as other state
residents. These regulatory changes were not supposed to affect the ability of needy
refugees to receive benefits, although this is not the way they were perceived by the
refugees themselves. What the programmatic changes essentially meant was that states
and local govemments would have to absorb a greater share of the fiscal responsibility for
refugee resettlement.
Until April 1, 1982, RCA and RMA had been available to eligible refugees during
their first 36 months in the U.S. In April 1982 a further reduction in the time limitation
for federal reimbursement to states was introduced in the form of a new policy goveming
the availability of cash and medical assistance. This policy, issued as an interim rule,
affected only refugees ineligible for AFDC, SSI, and Medicaid, for whom availability of
RCA and RMA was restricted to the first 18 months in the country. During their next 18
months, the refugees had to qualify under an existing state or local General Assistance program
(GA) on the same basis as other residents. ORR, however, would still reimburse
states for the cost of GA during the second 18-month period. At the same time, the period
during which refugees could benefit from the waiver of the family composition requirements
for cash assistance was reduced from 36 to 18 months. These programmatic
changes were expected to reduce the likelihood of unnecessary welfare dependency among
refugees facilitated by the certainty of extended periods of special support. Perhaps more
importantly, they were intended to defuse allegations of favoritism toward the refugees,
and, by implication, of inequitable treatment of other low income populations (IJ.S.
Department of Health and Human Services 1983:8).
The Refugee Assistance Amendments of 1982 reauthorized for one year (i.e., until
9/30/83) the legislation of the Refugee Act and the authorization of appropriations for
refugee assistance (U.S. Department of Health and Human Services 1983:18). To receive
assistance or reimbursement under the new legislation, states and welfare agencies had to
meet certain conditions designed to improve the refugee tracking system. In particular,
they had to provide assurance that they would notify the agency which initially resettled a
refugee when that refugee applied for cash or medical assistance. Additionally, refugees
who were full-time students in institutions of higher education were ruled ineligible for
cash assistance (U.S. Department of Health and Human Services 1983:9).
There is little doubt that the cumulative effect of these changes has not been
beneficial to the refugees' short-term well-being. In the first place, the termination from
RCA after 18 months for single refugees and couples with no minor children still needing
cash assistance meant that all benefits stopped unless a GA program for which they could
qualify was available locally. Then, even when GA is available, its benefit formula is usually
less generous than that for RCA. In addition, these further changes had the potential
to affect a substantial proportion of the refugee population: because of the high admission
levels of FY 1979 and FY 1980, wholly forty percent of the Indochinese refugee population
had entered the country between March 1979 and September 1980, and had therefore
been in the country for more than 18 months but less than 36 at the time of implementation.
So far, there has been little evidence to indicate that recent programmatic changes in
federal regulations have had much success in avoiding the development of an entitlement
mentality among refugees. This is largely because later arrivals, who overall have less
education and English competency than earlier arrivals, simply need more time to reach
self-sufficiency. However, there is some evidence that programmatic changes, by making
geographic mobility a necessary condition of economic survival for refugees who are not
yet self-sufficient, have contributed to the development of a peripatetic mentality, creating
54 Desbarats
Refugee Resettlement
hardship among the refugees and exacerbating their disproportionate impact on a few states
and counties of destination.
State Assistance Policies
Since the passage of the Refugee Act, federal resettlement assistance to refugees has
been provided by ORR through a state-administered refugee resettlement program. But,
states vary widely, not only in the structure of their public assistance programs and their
specific regulations and practices, but also in the organizational structure and administration
of their Refugee Programs (Berkeley Planning Associates 1982).
Benefit Availability. Even though needy refugees are theoretically entitled to the
same benefits whatever their place of residence, in practice, some categories of refugees
may find themselves eligible in some states and not in others. Before the Refugee Act,
refugee programs had not always been available in all states. For instance, the funding
interruption of the IRAP program between September 1977 and March 1978, caused by
delays in new appropriations, led eleven states to suspend temporarily their participation in
the refugee assistance program. The state of Missouri subsequently decided not to reinstate
its [RAP program (Strand and Jones 1985:37). More importantly, equity considerations
have led to the structuring of the Refugee Assistance Program along the same lines
as mainstream assistance programs, with the result that interstate disparities in the availability
and financial level of certain types of benefits affect the refugee population in practically
the same way as they affect the general population.
The 36-month federal reimbursement limit enforced after April 1981 introduced yet
another source of interstate differentials in benefit availability for time-expired refugees.
This stems from the presence or absence in the state of an AFDC-UP (Unemployed Parent)
program, i.e., an AFDC program that provides assistance to children in families with an
unemployed father. Because the AFDC-UP program has eligibility requirements to be
satisfied beyond those applying to AFDC-FG program, it is more restrictive than the general
program. In 1975 only 27 states had an AFDC-UP program (Hosek 1982:143). In the
other states, the AFDC program is not available if both parents are present and ablebodied,
a fact that has had some bearing on refugee settlement patterns. It is not surprising
that states without an AFDC-UP program should have become a source of secondary
migrants to those states with such a program. In California, for instance, most of the
refugees on public assistance are on the AFDC-lUP program, which covers two-parent families
with minor children, and entails no obligation for the principal wage-earner to accept
a job if it pays less than the amount received from assistance (Bach 1985:53-54). On the
other hand, in a number of refugee-receiving states without an AFDC-UP program or a GA
program, large numbers of refugees lost eligibility for assistance altogether as a result of
the March 1982 regulatory change.
Benefit Levels. The amount of assistance refugees receive under either RCA or
AFDC is determined by state government regulations for mainstream assistance programs.
Each state legislature has the power to "decide what need is and to what extent it is willing
and able to meet that need," and decides accordingly who is eligible for aid and how
much aid eligible persons can receive (Platky 1977:17). The great degree of state autonomy
in implementing the U.S. income-transfer system means that some states have more
generous assistance programs than others. Thus, a major consequence of the federally mandated
role of the states in the resettlement program has been to subject the level of refugee
assistance to the same state-to-state variations as regular AFDC programs. In 1975, for
instance, monthly AFDC benefit levels per recipient varied between an average of $14 in
Mississippi and $121 in Massachusetts. As of 1981, California has held the first rank
55 Nos. 65-66
KAS Papers
among all the conterminous states in terms of AFDC benefit levels (U.S. Social Security
Administration 1975, 1982). As a result, eligible refugees in different states receive substantially
different amounts, despite Congress' attempts to mitigate the wide variations in
benefit levels through the food-stamp program. If, as it has been suggested, the size of the
benefit check is more important than eligibility policies as a pull factor for the refugees
(North 1985:47), the potential consequences of these disparities for the geographic distrbution
of the refugees population can be easily anticipated. There is no dearth of empirical
evidence documenting the effect of state welfare levels on the migration patterns of disadvantaged
Americans (De Jong and Donelly 1973; Cebula 1979).
Local Assistance Policies. County assistance policies essentially affect single
refugees and childless couples, who are not eligible for AFDC. Refugees who are not eligible
for AFDC and who have not reached economic self-sufficiency by the time they
time-expire for the RCA program may qualify for county-funded and -administered GA
programs on the same basis as other county residents. Similarly, AFDC beneficiaries who
have not reached self-sufficiency by the time they lose eligibility can fall back on a GA
program. But the fact that not all counties administer GA programs introduces an additional
complication in the spatial structuring of the opportunities to which refugees
respond.
More recently, there has also been some evidence that counties receiving targeted
assistance funds, aimed at enhancing the employment opportunities of dependent refugees,
have attracted refugees from other counties. Similarly, the provision of better educational
programs specifically targeted for refugee populations has exercised a pull effect and the
availability of good mainstream vocational programs in community colleges has attracted
some refugees ineligible for refugee-specific programs.
While it cannot be denied that ethnic concentrations, employment prospects, and
wage levels probably have a significant influence in shaping refugee settlement patterns, it
is also certain that policy decisions affecting assistance availability, benefit levels, and
other refugee programs can become highly significant, at least in the short run. Ultimately,
the structure of opportunities determined by the interplay among the above factors
affects the conditions of refugee post-resettlement mobility at various geographic scales.
Overall, these factors seem to have compounded one another's effects to bring about
a marked increase in the regional concentration of the Indochinese refugee population,
whose distribution became more dissimilar from that of the general population. The proportion
of the refugee population that would have had to be redistributed for its distribution
at the state level to coincide exacdy with that of the American population increased
from 26 to 40 percent between 1975 and 1981 (Desbarats 1985:526). The fact that the
spatial concentration of the refugee population grew faster than that of initial placements
suggests that secondary migration also contributed to the increasing clustering.
Policy Influences on Refugee Migration Patterns
Secondary migration has been a common reaction of Indochinese refugees to the
federal dispersal policy. It has also been highly responsive to the evolution of assistance
policies over time. Refugees are extremely knowledgeable about federal and state assistance
policies and have used this knowledge in choosing their post-resettlement destinations.
Thus, while programmatic changes have often been a response to the evolution of
refugee settlement patterns, they have also had a major impact on these patterns, through
the way in which they have shaped incentives to secondary migration. The last part of the
paper presents specific examples of the ways in which the legislative context has
influenced refugee migration pattems.
56 Desbarats
Refugee Resettlement
Migration Levels
Indochinese refugees form a highly mobile population. Not only do they move a lot
more than the general U.S. population, but they also tend to move over longer distances.
By 1980, 45 percent of the 1975 arrivals were no longer living in their state of initial
resettlement (Baker and North 1985). Analysis of data from the Alien Registration Program
of the Immigration and Naturalization Service (INS) further indicates that among the
1975 to 1980 arrivals, one out of four had moved to a different state by January 1981, and
that, of those who arrived between October 1979 and December 1980, one out of five lived
in a different state in January 1981, less than fifteen months after they entered the country
(Gordon 1984:6). This high rate of mobility continued into the 1980s, as indicated by the
tabulation of data from the social security numbers of assisted refugees who entered the
U.S. between July 1980 and June 1983. By the end of the time period, 24 percent of them
had moved across a state line (Gordon 1984:10).
Although these various data sets refer to time frames that are not strictly comparable,
and to somewhat different subpopulations of refugees, the results concur in several
conclusions. Within their first few years in the U.S., refugees move at a higher rate than
other residents, and their propensity to move over long distances is higher than that of the
U.S. population as a whole (Gordon 1984:11). The majority of secondary moves occur
soon after resettlement, in the first two or three years of the refugees' initial adjustment
period. But with the passage of time, refugees tend to become more stable and their propensity
to move declines (Gordon 1984:6). These unusually high early mobility rates have
been attributed to the weakness of the refugees' ties to their places of initial resettlement,
their prior experience of international migration, and their status as new job seekers
(Forbes 1985a:9).
To some extent, refugee secondary moves reflect the same types of migration selectivity
that are found to operate for other populations, in that younger refugees, male
refugees, and those who are in the labor force are generally more mobile. But other forms
of selectivity specific to the refugee population also appear. The Sino-Vietnamese, for
instance, are more stable than the ethnic Vietnamese, a possible carryover of their greater
degree of residential stability in Vietnam (Desbarats 1986). Furthermore, among the 1975
entrants at least, refugees who had spent only a short time in processing camps were found
to be more stable residentially, as were those who had a sponsor, those who had friends in
the U.S. prior to arrival, those resettled in areas with large Vietnamese communities, those
resettled by Church World Service and Lutheran Immigration and Resettlement Services,
and-not unexpectedly-those who had initially been resettled in the state of their choice
(Baker and North 1985). After a variety of factors were combined into an index measuring
the difficulty of the resettlement experience, it appeared that the more difficult this
experience had been, the more likely a refugee was to move (Baker and North, 1985).
Migration Patterns
Limited inferences about reasons for moving may be made from information about
the origins and the destinations of secondary migrants. Geographic patterns will be considered
separately for the pre- and post-1980 periods because of differences in the data
sources available for the two periods, and because of changes in both levels and pattems
after 1980. In particular, the degree of spatial polarization of secondary migration patterns
has increased over time (Desbarats 1985:528).
Nos. 65-66 57
KAS Papers
In the first time period (1975-1980), movement was mostly in the direction of Southern
and Western states. From 1978 to 1980, 32 states had a consistently negative migration
balance for each of the three years, and another five had a negative balance for two of
the three years considered (Desbarats 1985:529). Even though, in a crude fashion, refugee
migration patterns could be said to be similar to those of the U.S. population, the degree of
geographic polarization of the refugees' post-resettlement moves far exceeded that of the
general population. Between 1975 and 1980, only three states-California, Washington,
and Kansas-had a consistently positive migration balance and another nine had positive
balances for two of the three years considered (Desbarats 1985:529). In contrast, for the
general U.S. population, 33 states registered net migration gains during the 1970s (U.S.
Bureau of the Census 1980). Of the states that continuously attracted refugees, California
received over two-thirds of refugee interstate migration-as against nine percent only of
general interstate migration (Gordon 1984:13). The first outmigration rank held by the state
of Missouri, which lost over one-third of its refugee population, reflects the effect of the
uncertainty over IRAP funding on the state's unwillingness to resume its participation in
the program.
In 1978, the most significant factors accounting for net rates of interstate migration
were average length of winter and average per capita income. While these two factors
remained significant in 1979, AFDC benefit levels and density of Asian immigrants also
contributed additional explanation of the migration patterns. By 1980, the effect of
climatic amenities on secondary migration had lost its significance, as had the presence of
an Asian immigrant population base. But additional explanatory factors emerged, such as
manufacturing wages, metropolitan population, and the proportion of refugees on assistance,
a surrogate measure for ease of eligibility. For all three years considered, AFDC
benefit levels were significantly and positively correlated with net migration rates (Desbarats
1985:533-34). Furthermore, of the specific states that had mostly positive migration
balances during the period considered, all but one-Texas-had above-average benefit levels,
and in nine of them, benefit levels were well above average.
As of 1980, patterns of secondary migration are more difficult to document systematically
because of the phasing out of the INS registration program, whose data provided a
baseline for the estimation of net migration. An ORR analysis of arrivals from October
1979 to December 1980 found that by January 1981, California had attracted 36 percent of
the known secondary migrants, again, a much higher proportion than for the population as
a whole. Only four other states-Rhode Island, Washington, Minnesota, and Oregon-had
positive migration balances. Again, these five in-migration states provided well aboveaverage
welfare benefits. Texas, the greatest loser, which ranked near the bottom nationally
in average AFDC levels (U.S. Social Security Administration 1981), lost 28 percent of
its original refugee arrivals over the period studied, gaining less than 5 percent of total
known migration (Gordon 1984:6).
A still more polarized picture emerges from the analysis of the interstate migration
matrix based on the Social Security numbers of assisted refugees who had been in the
country for less than three years as of the reporting date of June 1983. Five states had
positive migration balances-California, Massachusetts, Virginia, Rhode Island, and
Wisconsin-and California received 64 percent of all reported in-migration, with the state
of Washington the second favored destination. Out-migration was not as focused as inmigration,
and again, Texas was the major losing state, with 13 percent of all out-migrants.
The most important flows between specific origins and destinations for that period were
from Texas, Oregon, Washington, New York, Illinois, and Minnesota to California. In 16
states a much higher than average proportion of arrivals left the state and sought assistance
in another state. Most of these states have more restictive welfare eligibility criteria
58 Desbarats
Refugee Resettlement
and/or lower benefits, and/or no GA programs. The six states in which a much lower than
average proportion left to seek assistance elsewhere (California, Colorado, Massachusetts,
Minnesota, Michigan, New Jersey) generally have liberal cash assistance policies and high
benefit levels (Forbes 1985a 15). Overall, the interstate migration patterns of the refugees
appeared less diffuse and less local than those of the U.S. population (Gordon 1984:9-13).
Thus, while it has been stated that post-1980 patterns of secondary migration "agreed in
fundamental ways with the interstate migration patterns of other U.S. residents" (Forbes
1985a:9) this, again, is only true in a very superficial way.
The data base for the Social Security matrix is based on service recipients and is
therefore heavily biased toward assisted refugees. For this reason it cannot be taken as
representative of the refugee population as a whole. However, the difference in the degree
of migration polarization indicated by the INS data and the Social Security data confirms
that the attraction of California disproportionately affects dependent refugees, who generally
show much higher levels of migration to states with liberal assistance regulations
and high benefit levels.
Some Effects of the 1982 Refugee Act Amendments
The impact of the changes in RCA and RMA eligibility required by the 1982
Amendments to the Refugee Act provide a striking illustration of policy influences on
refugee migration patterns. When, in April 1982, the waiver of the family composition
rule for RCA and RMA became subject to an eighteen-month time limit, refugees who had
not yet reached economic self-sufficiency were left with the option of applying for existing
county-funded GA programs, if those were locally available. As of early 1982, needy
refugees who were about to reach time-expired status in states and counties where GA programs
were not available began to move out in anticipation of the regulatory changes.
The impact of these changes on out-migration rates is documented in two studies conducted
by the states of Washington and Oregon, which had consistently positive migration
balances until 1982. In Washington, where approximately 10,000 out of 30,000 refugees
were terminated from assistance, 25 percent of those affected by the new regulations had
moved out of state by August 1982 (State of Washington Department of Social Services,
1983). In Oregon, fifteen percent of the total population of 18,000 moved in anticipation
of the programmatic changes. The outflow was not composed exclusively of refugees who
had not become self-sufficient by the time they time-expired, however. It also included
some refugees previously employed in entry-level jobs without insurance benefits who had
quit their jobs to go on welfare elsewhere when their eligibility for RMA was cut off.
In both states, outmigration disproportionately affected upland Laotian refugees.
Thus, 90 percent of Oregon's upland Laotians left the state between February and July
1982, as against ten percent of the Vietnamese, even though this group was subject to the
same regulations. In Washington, 37 percent of the upland Laotians left, but only six percent
of the Sino-Vietnamese did so (Forbes 1985a 15). Officials from Catholic Charities in
Spokane told the Los Angeles Times that the city's Vietnamese population had dropped by
an estimated ten percent and the upland Laotian population by one-third. Two plausible
reasons may be advanced to explain why upland Laotians found themselves more vulnerable
to the reduction in the assistance period. In the first place, their nral backgrounds
much reduces the likelihood that they will reach economic self-sufficiency within eighteen
months of arrival, stretching their average period of dependency on cash assistance. In
addition, their dependency on cash and medical assistance is reinforced by the large size of
their families, and an average number of dependents which is much higher than for the
other Indochinese groups.
Nos. 65-66 59
KAS Papers
The majority of the out-migrants from Oregon and Washington went to California,
presumably attracted by that state's more liberal welfare system. It was around that time
that welfare officials in California complained that the state of Oregon had advised 5,000
jobless Indochinese refugees to move to California where they could collect welfare.
According to the director of the California Department of Social Services, refugees in Oregon
who called a telephone hotline were told: "If you want a job, go to Oklahoma City or
to Houston. If you want to go to school or go on welfare, go to California." He added
that the Oregon hot line (which had since been changed) even offered the refugees travel
money (New York Times May 2, 1982:13).
In Texas, one effect of the implementation of the 18 month cut-off limit was to
compound the employment market shrinkage caused by the 1981 recession. The inability
of laid-off refugees to fall back on GA programs-which Texas lacks-and their unwillingness
to rely on an AFDC program that has the second lowest benefit levels in the
nation turned Texas around from a net in-migration to a net outmigration state. More
recently, however, two seemingly contradictory trends have been noticed in the migration
patterns of refugees who were receiving cash assistance while enrolled in training programs
and who are seeking employment after completing their training. Employment service
providers now report serving refugees who have moved to Texas from states with
high welfare dependency rates, California in particular. In addition, there is some evidence
of return migration from states with liberal cash assistance policies to states with
more restrictive programs. Some officials, speculating on the motives for the behavior
underlying these unexpected migration patterns, have hypothesized that refugees may consider
some states to be good training grounds to learn English and obtain some skills, even
though they may not intend to live there permanently. When their training is completed
and they feel ready to enter the job market, they may then move to states with good
employment prospects (Forbes 1985a: 16).
Conclusion
In this paper I have used both direct and circumstantial evidence to document the
influence of domestic refugee resettlement and assistance programs on the migration and
settlement patterns of the Southeast Asian refugee population. It should be clear by now
that federal attempts at dispersing the refugee population through initial placement policies
have been undermined by two distinct but interrelated processes: 1) the refugees' own
desires for clustering, a sociocultural process which may be said to be partly economic
because of the refugees' heavy reliance on ethnic networks to locate employment opportunities;
and 2) federal assistance policies, which, by shaping economic incentives, further
encouraged clustering in a few states and localities. While initially the increase in the
geographic concentration of the refugee population was largely accomplished through
secondary migration, the process fed back onto itself because of the locational implications
of the family reunification policy. In view of the significant role of domestic refugee policies
on settlement pattems, then, it is worthwhile to reexamine the direct and indirect
consequences of refugee secondary migration for the refugees themselves and for the areas
that receive them.
When secondary migration occurs informally, as is most often the case, it can upset
long-range planning of resources, make health assessment more difficult, and limit the
Volags' ability to follow up on refugee progress. In addition, secondary migration means
renewed upheaval and additional expenses for the refugees themselves. But its effect on
tom19refug Sch t,f intstanc2&e,mbiat stsmt tombe semctAmive omloyd,t
1975 refugee cohort, for instance, migration seemned to be selective of thie unemployed,
601 Desbarats
Refugee Resettlement
and the earnings of those who had not moved were slightly higher than those of migrants.
On the other hand, there is some indication that migrants on average improve their
economic lot by migrating (Baker and North 1985), a finding confirmed by independent
evidence that migrants were doing better than non-migrants, and that those who had moved
to Texas and California, in particular, were more likely to be in the labor force as well as
more likely to be employed (Opportunity Systems Incorporated 1984).
Overall, the net effect of secondary migration has been to increase the concentration
of refugees in a small number of states, counties, and neighborhoods. It is probably fair to
say that, directly or indirectly, post-resettlement moves bear significant responsibility for
the disproportionate burden shouldered by states and counties with large refugee concentrations.
At the local level, refugee residential moves tend to promote increased concentration.
In Orange County, for instance, the proportion of a sample of refugees living in
"impacted" census tracts was found to be 28 percent before residential moves but 41 percent
afterwards (Desbarats and Holland 1983:37). Nationwide, the proportion of 1975
Indochinese entrants living in zip code areas with at least 3,000 refugees went up from 8.5
percent on arrival to 20 percent in 1980 (Baker and North 1985). The impact of increased
concentration at the local level-strain on social services, saturation of the housing stock,
increased visibility of the refugees, community resentment, and tense ethnic relations-has
been well documented.
The effect of geographic concentration at the state level is not as easy to summarize,
because secondary migration results in qualitative as well as quantitative changes in the
refugee population. While there is little doubt that much of the movement to Califomia,
for instance, has been precipitated by changes in federal regulations for cash and medical
assistance eligibility, it is probable that assistance policies also affect the composition of a
state's refugee population through selective in- and out-migration. Claims have often been
made that some state systems-such as the one in Texas-encourage economic selfsufficiency
and discourage assistance dependency (North 1985:48). But what has not been
acknowledged is the possibility that such systems do not so much encourage speedy
economic adaptation as the selective outmigration of refugees not yet ready to compete on
the job market.
In conclusion, refugee post-resettlement moves should be viewed as a rational
response to a changing pattern of opportunities for economic survival, and attempts at
dispersal should be viewed in a more balanced perspective, for having impeded the
refugees' efforts to become self-sufficient probably as much as for having helped them.
Such recognition would increase receptivity to the kind of policy recommendation recently
expressed by an experienced social worker, and quoted here as a final word.: "The problems
encountered by the refugees and those attempting to help them should not be complicated
by a policy which seeks to disperse the refugees and then ignores them when they do
what any rational observer would expect them to do-recongregate. Policies should be
geared toward congregation of refugee groups, not toward dispersal" (Yarling 1985:10).
References Cited
Bach, Robert
1985 Labor Force Participation and Employment of Southeast Asian Refugees in the United
States. In Aspects of Refugee Resettlement in the United States. U.S. Departnent of Health and
Human Services.
Nos. 65-66 61
KAS Papers
Baker, Reginald and David North
1985 The 1975 Refugees: Their First Five Years in America. Forthcoming. Washington, D.C.:
New Transcentury Foundation.
Bames, Thomas
1977 Of the 36 Alternatives: Indochinese Resettlement in America. U.S. Department of State,
Senior Seminar in Foreign Policy, Washington, D.C.
Berkeley Planning Associates
1982 Study of the State Administration of the Refugee Resettlement Program. Final Report.
Prepared for the Office of Refugee Resettlement, U.S. Departnent of Health and Human Services.
Cebula, Richard
1979 A Survey of the Literature on the Migration Impact of State and Local Govermnent Policies.
Public Finance 34:69-84.
De Jong, Gordon and William Donelly
1973 Public Welfare and Migration. Social Science Quarterly 54:329-44.
Desbarats, Jacqueline
1985 Indochinese Resettlement in the United States. Annals of the Association of American
Geographers 75:522-38.
1986 Edmic Differences in Adaptation: Sino-Vietnamese Refugees in the United States. Intemational
Migration Review 20.
Desbarats, Jacqueline and Linda Holland
1983 Indochinese Settlement Patterns in Orange County. Amerasia Jounal 11:23-46.
Forbes, Susan
1985a Residency Patterns and Secondary Migration of Refugees. Migration News 4:3-18.
1985b Adaptation and Integration of Recent Refugees to the United States. Washington, D.C.:
Refugee Policy Group.
Gordon, Linda
1984 Migration of Refugees Within the United States: New Data Sources and Findings. Paper
Presented at the Annual Meeting of the Association of American Geographers, Washington,
D.C.
Hosek, James
1982 The AFDC-Unemployed Fathers' Program: Determinants of Participation and Implications
for Welfare Reform. In Welfare Reform in America. Edited by P. Sommers. Boston: Kluwer
Nijhoff, 141-64.
International Sociological Association
1955 The Positive Contributions by Immigrants.
New York Times
1982 Oregon Told Refugees to Go to California. May 2:13.
North, David
1985 Refugee Eamings and Utilization of Financial Assistance Programs. In U.S. Department
of Health and Human Services. Aspects of Refugee Resettlement in the United States. Washington,
D.C.
Opportunity Systems Incorporated
1984 Eleventh Wave Survey of Southeast Asian Refugees. Washington, D.C.
Platky, Leon
1977 Aid to Families with Dependent Children: An Overview, October 1977. Social Security
Bulletin 40:17-22.
State of Washington Department of Social Services
1983 Refugee Assistance Termination Study and Ten-Month Follow Up. Olympia, Washington:
Department of Social Services.
Strand, Paul and Woodrow Jones, Jr.
1985 Indochinese Refugees in America: Problems of Adaptation and Assimilation. Durham,
N.C.: Duke University Press.
62 Desbarats
Refugee Resettlement
Teitelbaum, Michael
1980 Right Versus Right Immigration and Refugee Policy in the United States. Foreign Affairs
59:21-59.
U.S. Bureau of the Census
1983 State of Residence in 1975 by State of Residence in 1980. 1980 Census of Population
Supplementary Report, PC 80-S1-9.
U.S. Department of Health and Human Services. Social Security Administration. Office of Refugee
Resettlement.
1981, 1982, 1983 Refugee Resettlement Program, Report to the Congress. Annual. Washington,
D.C.: U.S. Government Printing Office.
U.S. Department of Health and Human Services. Office of Research and Statistics.
1982 Survey of the Social, Psychological, and Economic Adaptation of Vietnamese Refugees in
the U.S. 1975-1979. No. 13-11755. Baltimore.
U.S. Department of Health and Human Services. Social Security Administration. Office of Refugee
Affairs.
1979 Indochinese Refugee Assistance Program. Report to the Congress. Washington, D.C.:
U.S. Government Printing Office.
U.S. Social Security Administration
1975, 1981, 1982 Annual Statistical Supplement to the Social Security Bulletin. Annual. Washington,
D.C.: U.S. Government Printing Office.
Yarling, John
1985 Getting Rid of the Mousetraps. Paper presented at the Conference on Southeast Asian Refugees
in the United States: The First Decade. San Francisco, CA.
Nos. 65-66 63