DISCUSSIONPAPERSERIES
Forschungsinstitut
zur Zukunft der Arbeit
Institute for the Study
of Labor
Welfare Migration
IZA DP No. 6450
March 2012
Corrado Giulietti
Jackline Wahba
Welfare Migration
Corrado Giulietti
IZA
Jackline Wahba
University of Southampton
and IZA
Discussion Paper No. 6450
March 2012
IZA
P.O. Box 7240
53072 Bonn
Germany
Phone: +49-228-3894-0
Fax: +49-228-3894-180
E-mail: iza@iza.org
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in
this series may include views on policy, but the institute itself takes no institutional policy positions.
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center
and a place of communication between science, politics and business. IZA is an independent nonprofit
organization supported by Deutsche Post Foundation. The center is associated with the University of
Bonn and offers a stimulating research environment through its international network, workshops and
conferences, data service, project support, research visits and doctoral program. IZA engages in (i)
original and internationally competitive research in all fields of labor economics, (ii) development of
policy concepts, and (iii) dissemination of research results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be
available directly from the author.
IZA Discussion Paper No. 6450
March 2012
ABSTRACT
Welfare Migration*
This chapter reviews and discusses major theories and empirical studies about the welfare
magnet hypothesis, i.e. whether immigrants are more likely to move to countries with
generous welfare systems. Although economic theory predicts that welfare generosity affects
the number, composition and location of immigrants, the empirical evidence is rather mixed.
We offer possible explanations for the existence of such mixed evidence and highlight that
the literature so far has overlooked the presence of different migration regimes, as well as the
possibility of reverse causality between welfare spending and immigration.
JEL Classification: H53, J61, J68
Keywords: immigration, welfare spending
Corresponding author:
Corrado Giulietti
IZA
P.O. Box 7240
53072 Bonn
Germany
E-mail: Giulietti@iza.org
*
Forthcoming in “International Handbook of the Economics of Migration”. The authors are grateful to
an anonymous referee and to the editors Amelie F. Constant and Klaus F. Zimmermann for valuable
suggestions and comments on an earlier draft.
1 Introduction
For a long time economists have analyzed the consequences of immigration in the host countries,
and they have always been interested in why migration occurs and what its determinants
are. Yet while there is a well-established body of literature focusing on the push and pull factors
of immigration, such as wage differentials, macroeconomic conditions and social networks (see
Mayda, 2010 for a survey), only recently has the topic of “welfare migration” – i.e. whether
immigrants are more likely to move to countries with generous welfare systems – generated substantial
interest among scholars (a seminal work being Borjas, 1999; see also the forthcoming
2012 special issue of the International Journal of Manpower on “Migration, the Welfare State,
and European Labor Markets”).
At the same time, however, public worries about welfare migration have been growing. In recent
years a controversial debate has erupted on whether immigrants are exploiting the welfare
system. The concern is that immigrants move to countries with generous welfare and receive
social benefits without sufficiently contributing to the system. As an example of this widespread
perception, the 2009 Eurobarometer – a survey which monitors public opinion in the European
Union – reports that as many as 51 per cent of the individuals in the sample believe that immigrants
do not contribute to taxes as much as they benefit from social services (such as health)
and welfare programs (European Commission, 2010). Despite the intrinsic caveats implied by
the use of opinion surveys – e.g. public opinions towards immigrants are somewhat influenced
by the business cycle (Boeri, 2010) – this figure is symptomatic of the widespread perceptions
about welfare migration. (For further reference about public opinions on immigration, see the
chapter on attitudes towards migrants in this volume).
This chapter contributes to the debate on this topical question in two ways. First we review
and discuss major theories and empirical studies on the welfare magnet hypothesis; and second
we critically assess how the literature has developed and what the major challenges for future
research in this topic are.
The chapter is organized as follows: in Section 2 we provide an overview of welfare and immigration
in a selected group of Member States of the Organisation for Economic Co-operation
and Development (OECD). In Section 3 we review major theoretical works that model the relationship
between welfare and immigration. These studies predict that welfare programs affect
the number, composition and location of immigrants. We then survey two strands of the empirical
literature on the welfare magnet hypothesis in Sections 4 and 5, where we first examine
studies on whether immigrants are more likely to be welfare users than natives and then review
studies on whether immigrants choose countries or regions with generous welfare systems. The
empirical evidence is rather mixed. Several studies have found no evidence that welfare attracts
immigrants; others document the existence of a welfare magnet effect – albeit the economic
impact is often moderate. We outline possible reasons behind such contrasting evidence and
emphasize that the majority of the studies have overlooked the existence of different migration
regimes, as well as the possibility of reverse causality between welfare spending and immigration.
We conclude in Section 6.
2
2 Immigration and welfare: patterns and trends
2.1 Social expenditure patterns
Welfare refers to public programs with “social transfers” or “social benefits” (Adema and
Ladaique, 2009). In general it is possible to define two broad types of benefits: contributory and
non-contributory. The former includes unemployment insurance, which is compulsory in most
OECD countries, and old-age public pensions (in countries where there is a public provision).
The latter comprises of both universal income support programs, conceived as the financial
assistance of last resort (e.g. social assistance) and means-tested programs designed to reduce
the impact of vulnerable economic and social conditions (e.g. unemployment assistance and
housing benefits).1
Eligibility and coverage of these benefits vary from one country to another.
The standard approach in the welfare migration literature is to consider measures that capture
welfare state generosity. While several measures have been considered in the literature, the
descriptive statistics presented in this section are based on social expenditure in per cent of
gross domestic product (GDP), as this is thought to capture well the extent of resources that
a government devotes to social protection and it is also a measure which is fairly comparable
across countries and over time.2
A snapshot of welfare generosity is outlined in Figure 1, where the social expenditure and its
components in OECD countries with the highest immigration incidence are depicted for 2007,
the most recent period for which data are available. The figure provides a picture of welfare
expenditure and includes major areas of spending (both cash and in-kind benefits). These areas
have been defined as in Adema et al. (2011) and consist of: pensions, income support, health
and other expenditure. Figure 1 shows that there is substantial variation in social expenditure
across countries. The highest social expenditure is found in France, where spending is nearly
twice as high as that of Australia, Ireland and the United States – the three countries with the
lowest welfare spending in 2007.
There are also substantial differences in welfare components. Italy has the largest relative
spending on pensions, absorbing nearly 60 per cent of the total resources. Pension spending is
also particularly high in other Southern European countries, such as Greece and Portugal, but
also in Austria and France. These high levels reflect both a different mix of welfare policies
as well as the particular demographic trends of these countries. Spending on income support
– which includes unemployment insurance benefits as well as active labor market policies – is
relatively higher in Scandinavian countries. Countries with larger economies such as Canada,
1
For a more detailed classification see: Adema and Ladaique, 2009
2
A recent study by De Giorgi and Pellizzari (2009) investigates the welfare migration hypothesis,
focusing on an indirect measure of spending, the net replacement rate (NRR), defined as the share of
working income that is transferred, through unemployment benefits, to individuals who become unemployed.
Another possibility to classify welfare generosity would be to look at the duration of the benefits
or at their eligibility criteria. Nevertheless it transpires that these alternative measures are highly correlated
with spending. As an exercise, we have accessed OECD data about a particular type of welfare –
unemployment insurance – and obtained measures about total spending, NRR and durations. Comparing
data for 27 countries, we find a correlation of 0.53 between spending and NRR and a correlation of
0.61 between spending and durations.
3
Figure 1: Social expenditure and spending components in per cent of GDP, 2007
0
5
10
15
20
25
30
FRA
SWE
AUT
BEL
DNK
DEU
ITA
FIN
PRT
ESP
GRC
NOR
LUX
GBR
NLD
CHE
NZL
CAN
IRL
USA
AUS
%
GDP
Other expenditure
Health
Income support
Pensions
Source: SOCX database (http://stats.oecd.org), own elaborations. Pensions include spending on “old age”
and “survivors”. Income support includes spending on “incapacity”, “active labor market policy” and “unemployment”.
Other expenditure includes spending on “family”, “housing” and “other social policy areas”.
ISO country codes are: AUS=Australia, AUT=Austria, BEL=Belgium, CAN=Canada, CHE=Switzerland,
DEU=Germany, DNK=Denmark, ESP=Spain, FIN=Finland, FRA=France, GBR=United Kingdom,
GRC=Greece, IRL=Ireland, ITA=Italy, LUX=Luxembourg, NLD=Netherlands, NOR=Norway, NZL=New
Zealand, PRT=Portugal, SWE=Sweden and USA=United States.
Italy and the United States allocate a small share of the GDP to this type of welfare program.
Expenditure on health is relatively high in some Western European countries, such as Belgium,
France and Germany. Nevertheless, relatively high figures are also recorded for Canada, New
Zealand and the United States, where expenditure on health accounts for about 40 per cent of
total spending.
The remaining expenditure components mainly comprise of social assistance schemes, including
housing and family benefits, as well as unemployment assistance. The countries with largest
social assistance spending are Denmark, Sweden and the United Kingdom. Strikingly, countries
such as Italy and the United States devote a very marginal share of GDP to these programs.
Using social expenditure in per cent of GDP to describe time trends in welfare generosity, Figure
2 reports social expenditure patterns for major immigration countries aggregated into three
groups (EU-15/CHE/NOR, US and AUS/CAN/NZL).3
As the figure shows, spending increased
in all OECD countries from 1980 to 2010. After a sharp rise in the beginning of the 1990s, however,
expenditure remained constant. Interestingly, spending patterns of the EU-15/CHE/NOR
region and the United States were rather similar. However, the determinants of such changes
were somewhat different between the two regions. An inspection of expenditure components
reveals that while most of the increase in the EU-15/CHE/NOR region is attributable to pen-
3
The EU-15 comprises of Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland,
Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden and the United Kingdom.
4
sions, in the United States the area that contributed the most to the increase was health. In the
following we focus on immigration trends and the skill composition of immigrants and examine
whether immigration has evolved in the same manner as welfare spending.
Figure 2: Social expenditure in percentage of GDP over time
10
12
14
16
18
20
22
24
26
28
30
1980
1985
1990
1995
2000
2005
%
GDP
OECD - Total
US
EU-15/CHE/NOR
AUS/CAN/NZL
Source: SOCX database (http://stats.oecd.org), own elaborations.
2.2 Recent immigration patterns
From 1990 to 2010 26 million people migrated to Europe compared with 23 million to the
United States, 1.6 to Australia and 0.5 to Japan (Boeri, 2010). Countries such as Ireland and
Spain experienced a doubling of their foreign to domestic population ratios in the period 2000
to 2010. Table 1 reports most recent information about immigration flows.4
Overall, Western
Europe has attracted more immigrants than the United States. Within Europe Germany, Spain
and the United Kingdom had the largest number of immigrants, although the largest inflow
rates (i.e. normalized by population in the host country) are found in Australia, Austria and
Spain.
The composition by origin varies substantially, even within the EU-15 Member States (which
alone accounts for more than 90 per cent of the current immigrant population residing in the
whole European Union). For example the majority of immigrants in France are from African
countries, while more than half of the immigrants in the United Kingdom are from Asia. For
the United States the two major sending regions are Asia and Central and South America. Asia
4
Throughout the chapter we use the term “immigrant”, bearing in mind that the definition varies
across countries. For example the United Kingdom defines immigrants using a definition based on country
of birth; in Germany the definition is based on nationality. For more information on the data used see
http://stats.oecd.org.
5
is also the source of the majority of recent inflows to Australia, Canada and Japan. The table
suggests that both historical ties, as well as different level of economic development, appear to
be the main factors behind the substantial cross-country differences in the composition of recent
immigration flows.
Table 1: Immigration inflows by region of origin, selected OECD countries, 2009
Country
Total Inflow
EU-15
Other North Central
Africa Asia Otherinflow rate Europe America & South
America
Austria 91.8 0.011 26.6 47.0 2.0 1.6 3.8 18.4 0.6
Belgium 102.4 0.009 30.0 32.7 3.2 3.5 17.5 12.6 0.5
Denmark 30.8 0.006 24.1 40.5 3.1 2.2 4.0 23.3 2.8
Finland 18.1 0.003 11.8 27.4 1.9 2.0 10.9 43.7 2.3
France 126.0 0.002 0.3 7.2 2.4 6.4 61.8 21.5 0.5
Germany 599.1 0.007 15.1 50.8 3.4 2.9 4.2 22.8 0.8
Italy 267.4 0.004 9.9 20.4 2.9 10.7 28.0 28.0 0.0
Luxembourg 14.5 0.029 72.6 12.6 2.2 2.4 4.6 5.1 0.6
Netherlands 104.4 0.006 28.4 25.5 3.7 4.2 6.3 20.0 11.9
Portugal 31.0 0.003 20.8 38.6 1.5 9.5 20.8 8.9 0.0
Spain 469.3 0.010 14.9 18.9 1.1 32.5 21.2 11.3 0.1
Sweden 82.8 0.009 17.2 22.7 1.9 2.9 15.6 37.2 2.5
UK 329.0 0.005 16.7 12.8 5.2 0.0 6.1 54.1 5.2
Australia 222.9 0.010 18.7 2.1 2.2 1.4 9.2 50.2 16.2
Canada 251.9 0.007 10.0 4.7 3.9 10.6 12.0 58.0 0.8
US 1129.7 0.004 3.9 4.6 1.4 40.8 11.2 37.3 0.6
Source: OECD International Database (http://stats.oecd.org). Flows for Italy and Denmark refer to 2008.
The inflow rate corresponds to immigration inflow over the total population in the country.
Yet there are substantial differences across countries in terms of the skill and education level
of immigrants. Hence it is important to also consider the patterns of immigration by education
level. The most recent and accurate estimates are available for 2001.5
Figure 3 depicts
how patterns of high- and low-skilled immigration have evolved during the period 1991-2001.
The top panel of the figure represents the stocks of both immigrants with primary education
(low-skilled) and those with tertiary education (high-skilled) in 1991.6
Stocks are represented
in the map in percentage of the total population. As can be seen, largest immigrant stocks
with primary education were found in Australia, Canada, New Zealand and regions of Western
Europe. In the United States, as well as in the remainder of Western Europe and Sweden, the
percentage of immigrants with primary education was between 2 and 4 per cent, while in all
remaining countries the relative stocks were below 2 per cent. In Australia, Canada, Japan and
the United States the stocks of immigrants with tertiary education were broadly similar in size
to those with primary education. A different picture appears in Europe, where in all countries
except Switzerland the percentage of highly educated immigrants was below 2 per cent.
The lower panel of Figure 3 depicts changes in the relative stocks of immigrants between 1991
and 2001. During this period overall immigration increased substantially; however the patterns
5
Migration estimates are obtained from the “Data Set 1990-2000 with Gender Breakdown (Rel.
2.1)” (http://perso.uclouvain.be/frederic.docquier/oxlight.htm). These estimates are obtained combining
sources such as censuses and population registers. For more details see Docquier and Marfouk (2006)
and Docquier et al. (2009).
6
The dataset also provides information on immigrants with secondary education, which are not considered
in the figure. Yet comparing primary and tertiary educated immigrants provides insight for
understanding the patterns of high- and low-skilled immigration during 1991 to 2001.
6
differed for the groups of immigrants with low and high education. While it is possible to
observe a strong association between the stock of immigrants with tertiary education and its
decennial change (the correlation for the countries in the map is 0.61), there seems to be a major
redistribution across countries of low-educated immigrants (the correlation between the 1991
stock and the 1991 to 2001 change is 0.10). Hence countries which had relative large stocks of
low-educated immigrants in 1991 have actually experienced a less than proportional increase in
the stock of immigrants with primary education.
There are many factors behind these different trends, including differences in how the composition
of skills across sending countries has evolved, but also changes in both pull and push factors
of immigration, such as welfare spending, macroeconomic conditions, and immigration policies
in both sending and receiving countries.
Figure 3: Immigration: relative stocks (1991) and changes in relative stocks (1991-2001)
Stocks of immigrants in % of population, 1991
Change in the stocks of immigrants in % of population (1991-2001)
Source: own elaboration from the “Data Set 1990-2000 With Gender Breakdown (Rel 2.1)”
(http://perso.uclouvain.be/frederic.docquier/oxlight.htm). Population data are from the OECD (http://stats.oecd.org.).
Boundaries data are from http://www.gadm.org/world. Upper panel: immigration stocks for Denmark, Finland,
France, Norway, Spain, Sweden, Switzerland and the United States refer to year 1990. Stocks for the Netherlands refer
to 1996. Stocks for the remaining countries refer to 1991. Lower panel: immigration stocks for Denmark, Finland, the
Netherlands, Spain, Sweden, Switzerland and the United States refer to year 2000. Stocks for France refer to 1996.
Stocks for the remaining countries refer to 2001.
An important question is whether welfare spending has acted as a magnet for immigration
and potentially altered the skill distribution of immigrants. Indeed when examining the pro-
7
portion of unskilled immigrants – defined as the share of those with primary education out of
the total stock of immigrants – and welfare – measured as social expenditure as a percentage
of GDP – Figure 4 shows a positive relationship. This might suggest a possible welfare magnet
effect.
The empirical studies that we review in the following sections investigate whether welfare generosity
in developed countries is actually one of the determinants of the changes in immigration
patterns and whether immigrants of certain skills have been more prone to move to destinations
with more generous social spending systems. Before surveying these empirical works, however,
we briefly outline major theories linking welfare and immigration.
Figure 4: Low-skilled immigration and social expenditure
AUS
AUT
BEL
CAN
DNK
FIN
FRA
DEU
GRC
IRL
ITA
LUX
NLD
NOR
NZL
PRT
ESP
SWE
CHE
GBR
USA
0.2.4.6.8
Shareofimmigrantswithprimaryeducation,2001
15 20 25 30 35
Social expenditure in % of GDP, average 1990-2000
Source: own elaborations from the “Data Set 1990-2000 With Gender Breakdown (Rel 2.1)”
(http://perso.uclouvain.be/frederic.docquier/oxlight.htm) and SOCX database (http://stats.oecd.org).
3 Theories on welfare and migration
3.1 The welfare magnet hypothesis
The welfare magnet hypothesis was first coined in a seminal paper by Borjas (1999). This
hypothesis refers to how welfare generosity acts as a pull factor for migration and how it influences
the skill composition of immigrants. More specifically, it states that immigrants prefer to
locate in countries with generous welfare provisions to insure themselves against labor market
risks. This effect may not be necessarily limited to unskilled immigrants, since also high-skilled
immigrants may prefer to live in countries with larger social benefit systems, e.g. because economic
fluctuations might affect their employment perspectives irrespective of the skill level.
8
Borjas (1999) argues that welfare could influence immigration through several channels. First
more generous welfare countries might attract immigrants who otherwise would have not immigrated.
Second the existence of social safety nets might also retain immigrants who would have
otherwise returned to their country of origin. To prove his point, Borjas uses the example of
the United States, claiming that substantial interstate dispersion in welfare benefits has affected
the residential location choices of international immigrants. Immigrants to the United States
are individuals who have already borne the cost of immigration. Hence the cost of choosing
one state or the other is negligible. This implies that immigrants who receive welfare tend to
choose, and thus be clustered in, states which offer the highest social benefits. On the other
hand, welfare will not be a strong incentive to migrate across states for US natives because they
still have to bear the costs of moving. As a consequence, native welfare recipients tend to be
more dispersed across states.
While it efficiently explains the potential role of welfare in attracting immigrants, Borjas’ model
ignores the importance of the role other determinants of immigration play, such as social networks.
Networks provide information about labor market opportunities and thus reduce the
cost of migration (see Beine et al., 2011). On the other hand they can also be a source of
information on welfare benefits for potential immigrants who are still in the source country.
Another important factor is immigration policy. In many instances, in particular when it comes
to destination choices, immigrants are restricted in their choices: they are not free to move
to the country with the most generous welfare system – even if they wished to. This may be
because of certain barriers to migration – such as language and physical distance – but more
importantly, as we discuss below, because of restrictive immigration policies.
3.2 Migration regimes, welfare and selective immigration
In a recent paper Razin and Wahba (2011) argue that the generosity of the welfare state
may affect the skill composition of immigrants, depending on the type of immigration policy
adopted. In a free-migration regime a typical welfare state with relatively abundant capital and
high total factor productivity (implying relatively high wages for all skill levels) attracts both
unskilled and skilled immigrants. On the other hand the generosity of the welfare state attracts
unskilled immigrants, as they expect to gain more from welfare benefits than what they expect
to pay in taxes for these benefits, i.e. they are net beneficiaries of the generous welfare state. In
contrast, potential skilled immigrants are deterred by the generosity of the welfare state. Thus
the welfare state would tilt the skill composition of immigrants towards the unskilled.
In the restricted-migration regime these same considerations lead voters to favor skilled immigration
and restrict unskilled immigration. Voters are motivated by two considerations: how
immigration affects their wages and how it bears on the finances of the welfare state. Typically
one would expect unskilled immigration to depress wages of (substitutable) low-skilled workers
and to increase wages of (complementary) high-skilled workers. The opposite is true of skilled
immigration. The effect of immigration on the finances of the welfare state is common to voters
of all skills, since more skilled immigrants are net contributors to the welfare state, whereas
9
unskilled immigrants are net beneficiaries. From a public finance point of view native-born
voters of all types of skills would therefore favor skilled immigration and support restrictions for
unskilled immigration. Hence the welfare state in a managed immigration regime would favor
skilled immigration.
To sum up, the welfare magnet hypothesis put forward by Borjas (1999) predicts that immigrants
are attracted by the generous welfare state. Razin and Wahba (2011) extend Borjas’ idea
and argue that the welfare magnet hypothesis is only expected in free-migration regimes, where
migrants are free to self-select and the generosity of the welfare state attracts mainly unskilled
immigrants. In a managed-migration regime, however, demand for immigrants would favor the
high skilled as net contributors to the welfare system.
4 Are immigrants more likely to be welfare recipi-
ents?
The theoretical frameworks presented suggest that welfare systems are potential pull factors
for international migration. More specifically, welfare systems influence migration decisions and
migration selection, potentially leading to an adverse selection of immigrants. This section
examines the empirical evidence on the welfare magnet hypothesis.
A number of papers address this hypothesis by asking whether immigrants are more likely to
be welfare recipients, i.e. they test the welfare dependency of immigrants.
4.1 Evidence from the United States
Using information on immigrant participation in the welfare system derived from the 1970
and 1980 US Censuses Studies, Borjas and Trejo (1991) show that recent immigrant cohorts use
the welfare system more intensively than earlier cohorts. In addition they find that immigrant
households “assimilate” into welfare the longer they have been in the United States. Further
evidence is provided by Borjas and Hilton (1996), using data drawn from the Survey of Income
and Program Participation (SIPP). They find that immigrants are more likely to receive cash
benefits than natives and that the immigrant-native gap increases when non-cash transfers are
included. Furthermore immigrants appear to receive welfare more frequently and for longer
spells than natives. Finally the authors find evidence that there is a correlation between the
type of welfare benefits received by earlier immigrants and those received by newly arrived immigrants.
This suggests that immigrant social networks may disseminate information about the
opportunity of receiving welfare benefits.
Borjas (1999) shows that immigrants who receive welfare benefits tend to concentrate in states
with higher benefits – contrary to natives. This is taken as evidence that immigrants are more
responsive to changes in welfare than natives. Yet Borjas recognizes that his findings are relatively
weak in terms of statistical significance.
Although the literature on the welfare magnet hypothesis initially focused on international migration,
many studies on the United States investigate welfare migration of natives across states,
10
obtaining mixed results. For example, Levine and Zimmerman (1999) find no substantial evidence
of welfare migration. Using the National Longitudinal Survey of Youth (NLSY) between
1979 and 1992, they examine the extent to which cross-state differences in welfare generosity is
associated with interstate migration. Their approach is to compare patterns of internal migration
of a group which is eligible to welfare receipt (poor, single women with children) with a
similar group, albeit not eligible to welfare (other poor households).
A similar approach is also implemented in other studies, where the migration behavior of single
mothers – a welfare-prone group – is compared with the migration behavior of groups less prone
to receiving welfare, such as married mothers. Gelbach (2000) argues that welfare migration
should occur mainly among mothers with young children. He provides evidence of the existence
of “life cycle welfare migration”. However, statistical results for the 1990 Census are somewhat
weaker than the evidence for the 1980 Census. Enchautegui (1997) explores the effects of several
determinants – including welfare payments – on women’s interstate migration. She finds that
welfare is positively associated with the probability of moving from state to state and that the
effect is larger for single mothers with young children and for women who have not recently
participated in the labor market.
A similar approach using the 1980 and 1990 Censuses is adopted by Meyer (1998). He compares
inter-regional migration of single mothers with moves by single women without children
and married mothers. He finds evidence of moderate welfare migration, particularly when the
sample of high-school dropouts is considered. More recently McKinnish (2005, 2007) provides
evidence of welfare migration by focusing on cross-border migrations (i.e. short-distance moves).
Her strategy is based on comparing the welfare use of individuals residing close to state borders
with that of individuals living in the inner part of the state. Although imprecisely estimated,
the results of her study indicate that welfare benefits affect the migration decision of individuals
more likely to receive welfare.
4.2 Evidence from Europe
The issue of welfare dependency of immigrants has also been examined in European countries.
Using panel data for Sweden for the years 1990 to 1996, Hansen and Lofstrom (2003)
compare the welfare use of immigrants and natives. Their findings indicate that after controlling
for observable characteristics, welfare participation is higher among immigrants than among natives.
However welfare use among immigrants tends to decrease with time spent in Sweden. This
“assimilation out of welfare” for Sweden somewhat contrasts with the findings of Borjas and
Trejo (1991) for the United States. Riphahn et al. (2010) investigate a similar research question,
focusing on Turkish immigrants using the German Socio-Economic Panel (SOEP). They show
that this particular group is more prone to welfare use than natives. However after controlling
for a set of individual and household characteristics, evidence of residual welfare dependency
is statistically significant only for second-generation immigrants. Barrett and McCarthy (2008)
provide a review of studies that compare immigrants’ and natives’ welfare use and conclude that
the evidence across countries is mixed. As further indication of such contrasting results, they
compare the case of the United Kingdom, where immigrants use welfare more intensively than
11
natives, and of Ireland, where instead natives are more likely to participate in welfare programs.
Boeri (2010) uses the EU Survey of Income and Living Conditions (EU-SILC) and finds evidence
that unskilled immigrants are net recipients of non-contributory benefits, particularly in countries
with generous welfare systems. Focusing on Italy Pellizzari (2012) combines data from the
EU-SILC (containing information on welfare use) with administrative data (containing information
on applications for locally provided welfare programs). His results show that immigrants
from outside the EU are more prone to apply for welfare benefits, although once controlling
for individual and household characteristics, such an effect decreases substantially. Pellizzari
discusses that geographical sorting is a likely explanation of such results, since immigrants tend
to self-select to wealthier regions of Italy, which are also the most generous in terms of welfare.
Evidence on Europe also comes from the recent enlargement of the European Union to include
Central and Eastern European countries (new Member States). While the majority of the old
Member States imposed a period of “transitional rules” – adopting restrictions on the immigrations
of workers from the new Member States – a few countries (Ireland, Sweden and the
United Kingdom) freely opened their labor markets to immigrants. This raised the question
of whether immigrants coming from relatively lower income countries would have “abused” the
welfare system of the receiving countries. Kahanec et al. (2009) summarize the findings of studies
which examine welfare access in Ireland and Sweden during the post-enlargement period. In
Ireland no evidence of excess welfare use by immigrants is found; in Sweden, while immigrants
are more likely to receive social assistance, they are less likely to participate in other welfare
programs such as unemployment or sickness benefits. Blanchflower and Lawton (2009) analyse
the special case of the United Kingdom, where the government imposed restrictions on welfare
access to immigrants from the new Member States. On the one hand the authors provide confirmation
that only few immigrants from accession countries obtained welfare benefits such as
the “Income Support” or the “Jobseeker Allowance”. At the same time, however, they show
that these immigrants exhibit a higher probability of being in work when compared to natives
and immigrants from other origins. This leads them to conclude that “[t]hey came to work and
not to claim benefits” (Blanchflower and Lawton, 2009, p.188). Similarly, Constant (2011), in
her discussion about the effect of EU enlargement, concludes that no evidence of excess welfare
use can be found. Further discussion about the consequences of EU enlargement is provided in
the chapter “EU Enlargement and the European Labor Markets” in this volume.
To summarize, studies about the United States provide somewhat more consistent evidence of
immigrants’ residual welfare dependency. These studies, however, exploit variation across US
states, where interstate mobility is relatively higher compared to mobility within Europe. Immigrants
in the United States are free to choose the state where to locate and can select the
one with most generous welfare benefits (Nannestad, 2007). On the other hand the evidence
that immigrants in Europe are more likely to be welfare recipients than natives is rather mixed.
This is not surprising given the difference in the types of benefits offered and the heterogeneity
in welfare eligibility criteria across European countries. Furthermore, if evidence of immigrant
welfare dependency is found, it appears to be rather small.
It is important to note that most of the studies cited (one exception being Borjas, 1999) ex-
12
amine the welfare dependency of immigrants as an indication for the welfare magnet. However
immigrants’ excess welfare use does not necessarily imply that generous welfare states attract
immigrants. In other words, studies that examine welfare dependency among immigrants (or
the welfare use gap between immigrants and natives) do not convincingly test the magnet hypothesis,
i.e. whether immigrants’ decisions to migrate and their destination choice are affected
by the generosity of the welfare systems. There might be several reasons behind immigrants
using welfare more intensively than natives. For example immigrants may have unobservable
characteristics that make them more prone to be on welfare. Furthermore welfare dependency
could be triggered by labor market discrimination in accessing jobs (see Barrett and McCarthy,
2008). A more direct approach to test the welfare magnet hypothesis comes from another branch
of literature, which we survey in the next section.
5 Is welfare a magnet for immigrants?
Several papers examine the welfare magnet hypothesis by focusing on the locational choice
of migrants. Southwick (1981) presents one of the earliest studies about welfare migration. He
uses data drawn from a US study about the Aid to Families with Dependent Children (AFDC)
program and presents several tests for the welfare migration hypothesis. He finds that migration
flows between regions with large differentials in terms of benefits consist mostly of women
who are AFDC recipients. Using information on the states of residence in 1975 and 1979 for
a sample of welfare recipients from the Panel Study of Income Dynamics, Gramlich and Laren
(1984) show that AFDC beneficiaries, although exhibiting very low interstate mobility, are more
likely to move to a welfare generous state than to a low-benefit state. More recent evidence
about Europe is provided by De Giorgi and Pellizzari (2009), who combine data from the European
Community Household Panel (ECHP) with information from the OECD Database on
Unemployment Benefit Entitlements and Replacement Rates. Welfare generosity in their paper
is measured using the net replacement rate (NRR), i.e. the ratio between the income received
when not working (e.g. unemployment benefits) and the average wage. They test the welfare
magnet hypothesis by considering immigration in the EU-15 and find that welfare generosity
influences migration decisions, albeit the effect is small. In a study about the determinants of
immigration flows to OECD countries Pedersen et al. (2008) find that while social networks are
an important pull factor for immigrants, welfare – measured by social expenditure in per cent of
GDP – does not exert a significant role in attracting immigrants. They argue that immigration
policies might have prevented the potential adverse selection of immigrants.
Focusing on the skill composition of immigrants, Br¨ucker et al. (2002) find that welfare-generous
countries attract low-skilled workers, whilst countries with low social spending are more likely
to be a magnet for high-skilled workers, since taxes are also low in these countries. As a consequence,
welfare generosity may induce a negative sorting of immigrants.
One important issue that is seldom addressed in the empirical literature on the welfare magnet
hypothesis is the endogeneity between welfare and immigration. A few studies have shown that
actually immigration might affect welfare generosity. Using OECD panel data for the period
13
1990 to 2001, B¨oheim and Mayr (2005) find that low-skilled immigration decreases public spending,
while high-skilled immigration produces the opposite effect. The recent work by Giulietti
et al. (2012) tackles the question directly of whether there exists reverse causality between welfare
and immigration. The welfare magnet hypothesis is explored in the context of a particular
program – unemployment insurance – and two potential sources of endogeneity are discussed.
First, immigrants might affect spending by directly influencing the spending on unemployment
benefits in per cent of the GDP (through participation in welfare programs, but also through
taxes and consumption). Second, welfare policy could react to increasing immigration, and
policymakers could encourage or discourage welfare participation of immigrants by intervening
in aspects such as eligibility criteria or welfare duration. To address the endogeneity issue,
the authors use the number of parties in the government coalition as an instrumental variable
for unemployment benefits. The rationale is that public sectors are larger when coalitions are
formed by a greater number of political parties. At the same time this instrument is thought
to be uncorrelated with immigration. Using a sample of 19 European countries over the period
1993 to 2008, the ordinary least squares estimates show that unemployment benefit is positively
correlated with immigration flows from non-EU countries, but not with inflows from EU origins.
However, instrumental variables and generalized method of moments techniques yield an essentially
zero causal impact of unemployment benefits on immigration inflows from both areas.
Another recent paper that also investigates the endogeneity of welfare generosity is Razin and
Wahba (2011). They control for the endogeneity of total social spending per capita and find
strong support for the magnet hypothesis under the free-migration regime (as represented by
migration within the EU), and for the “fiscal burden hypothesis” under the restricted-migration
regime (as represented by migration from outside the EU). Their results are robust to using
total social spending as a percentage of GDP and for correcting for differences in educational
quality as well as for returns to skills between the source and host countries.
To summarize, although empirical evidence on the welfare migration hypothesis is rather mixed,
there are at least two potential important factors behind these unclear results. The first is that
the majority of the studies above have ignored the endogeneity of the welfare system and immigration.
This might have produced biased results in the estimations. Immigration may
affect directly or indirectly the level of social spending, depending on many factors, such as the
skill level of immigrants, the composition of the immigrant households, their proclivity to be
in welfare programs, and also the duration and eligibility conditions of the programs. Indeed
several papers have modeled the potential influence of immigration on redistribution and welfare
spending (Dolmas and Huffman, 2004; Razin et al., 2011). The second reason behind the
mixed empirical findings might be the result of ignoring the immigration regime (i.e. whether
immigration is free or restricted), which tends to underestimate the implications of immigration
selectivity within the context of the welfare magnet hypothesis.
14
6 Conclusion and policy implications
According to the studies reviewed in this chapter, it is plausible to conclude that fears about
immigrant abuse of welfare systems are somewhat unfounded or at least exaggerated. Overall
the empirical evidence on the welfare magnet hypothesis is mixed. However when evidence of a
magnet effect is found, the impact tends to be rather exiguous.
We have explored two potential sources for the conflicting empirical results: the endogeneity of
welfare and immigration and whether immigration in the country is free or restricted. Recent
empirical evidence suggests that reverse causation between welfare and immigration potentially
exists. Thus further exploring the issue of reverse causality between immigration and social
spending constitutes a potential avenue for future theoretical and empirical work aiming to
test the welfare magnet hypothesis. Further research will also need to consider explicitly the
immigration policies and their implications.
It is also important to note that welfare is one of the many pull factors of immigration. Future
research should attempt to accurately quantify the role of welfare generosity in relation to other
factors, such as wage differentials, labor market conditions, tax systems and social networks.
Our review suggests that the number and characteristics of immigrants are potentially affected
by not only immigration policies – which are meant to directly affect immigration flows – but
also by other policies, such as welfare programs. Hence policymakers should be aware of the
interactions between immigration and welfare policies. One of the major findings of a recent
study by Zimmermann et al. (2012) is that while raw statistics show that welfare receipt is
higher among immigrants in most of the European Union, when controlling for socio-economic
characteristics, such welfare dependency persists in only a few Member States. This suggests
that characteristics of immigrants directly influenced by immigration policies – such as their
skill level – are important determinants of immigrants welfare use. Hence policymakers should
focus on the design of selective immigration policies and at the same time should intervene on
welfare programs attributes (e.g. contributory nature and eligibility criteria) by taking into
account a country’s immigration pattern and the characteristics of immigrants.
How well the two types of policies are integrated will have consequences on the important issues
which are at the core of current debate about immigration, such as the sustainability of the
welfare systems versus the potential of immigration to alleviate labor shortages and counteract
the effects of an aging population.
15
References
Adema, W., P. Fron, and M. Ladaique (2011). Is the European Welfare State Really More
Expensive? Indicators on Social Spending, 1980-2012; and a Manual to the OECD Social
Expenditure Database (SOCX). OECD Social, Employment and Migration Working Papers
124.
Adema, W. and M. Ladaique (2009). How Expensive is the Welfare State? Gross and Net
Indicators in the OECD Social Expenditure Database (SOCX). OECD Social, Employment
and Migration Working Papers 92.
Barrett, A. and Y. McCarthy (2008). Immigrants and Welfare Programmes: Exploring the Interactions
Between Immigrant Characteristics, Immigrant Welfare Dependence, and Welfare
Policy. Oxford Review of Economic Policy 24(3), 543–560.
Beine, M., F. Docquier, and C¸. ¨Ozden (2011). Diasporas. Journal of Development Economics
95(1), 30–41.
Blanchflower, D. G. and H. Lawton (2009). The Impact of the Recent Expansion of the EU on
the UK Labor Market. In M. Kahanec and K. F. Zimmermann (Eds.), EU Labor Markets
After Post-Enlargement Migration, pp. 181–218. Berlin: Springer Verlag.
Boeri, T. (2010). Immigration to the Land of Redistribution. Economica 77(308), 651–687.
B¨oheim, R. and K. Mayr (2005). Immigration and Public Spending. IZA Discussion Paper
1834.
Borjas, G. J. (1999). Immigration and Welfare Magnets. Journal of labor economics 17(4),
607–637.
Borjas, G. J. and L. Hilton (1996). Immigration and the Welfare State: Immigrant Participation
in Means-Tested Entitlement Programs. The Quarterly Journal of Economics 111(2), 575–
604.
Borjas, G. J. and S. J. Trejo (1991). Immigrant Participation in the Welfare System. Industrial
and Labor Relations Review 44(2), 195–211.
Br¨ucker, H., G. S. Epstein, B. McCormick, G. Saint-Paul, A. Venturini, and K. F. Zimmermann
(2002). Managing Migration in the European Welfare State. In T. Boeri, G. Hanson, and
B. McCormick (Eds.), Immigration Policy and the Welfare System, pp. 1–168. Oxford: Oxford
University Press.
Constant, A. F. (2011). Sizing It Up: Labor Migration Lessons of the EU Enlargement to 27.
IZA Discussion Paper 6119.
De Giorgi, G. and M. Pellizzari (2009). Welfare Migration in Europe. Labour Economics 16(4),
353–363.
16
Docquier, F., B. L. Lowell, and A. Marfouk (2009). A Gendered Assessment of Highly Skilled
Emigration. Population and Development Review 35(2), 297–321.
Docquier, F. and A. Marfouk (2006). International Migration by Educational Attainment, 1990-
2000. In C¸. ¨Ozden and M. Schiff (Eds.), International Migration, Remittances and the Brain
Drain, pp. 151–200. New York: Palgrave Macmillan.
Dolmas, J. and G. W. Huffman (2004). On the Political Economy Of Immigration and Income
Redistribution. International Economic Review 45(4), 1129–1168.
Enchautegui, M. E. (1997). Welfare Payments and Other Economic Determinants of Female
Migration. Journal of Labor Economics 15(3), 529–554.
European Commission (2010). Eurobarometer 71: Future of Europe.
Gelbach, J. B. (2000). The Life-Cycle Welfare Migration Hypothesis: Evidence from the 1980
and 1990 Censuses. University of Maryland, College Park Working Paper 00-01.
Giulietti, C., M. Guzi, M. Kahanec, and K. F. Zimmermann (2012). Unemployment Benefits
and Immigration: Evidence from the EU. International Journal of Manpower, Special Issue
on “Migration, the Welfare State, and European Labor Markets” (forthcoming).
Gramlich, E. M. and D. S. Laren (1984). Migration and Income Redistribution Responsibilities.
Journal of Human Resources 19(4), 489–511.
Hansen, J. and M. Lofstrom (2003). Immigrant Assimilation and Welfare Participation. Journal
of Human Resources 38(1), 74–98.
Kahanec, M., A. Zaiceva, and K. F. Zimmermann (2009). Lessons from Migration after EU
Enlargement. In M. Kahanec and K. F. Zimmermann (Eds.), EU Labor Markets After PostEnlargement
Migration, pp. 3–46. Berlin: Springer Verlag.
Levine, P. B. and D. J. Zimmerman (1999). An Empirical Analysis of the Welfare Magnet
Debate Using the NLSY. Journal of Population Economics 12(3), 391–409.
Mayda, A. M. (2010). International Migration: A panel Data Analysis of the Determinants of
Bilateral Flows. Journal of Population Economics 23(4), 1249–1274.
McKinnish, T. (2005). Importing the Poor: Welfare Magnetism and Cross-Border Welfare
Migration. Journal of Human Resources 40(1), 57–76.
McKinnish, T. (2007). Welfare-Induced Migration at State Borders: New Evidence from MicroData.
Journal of public Economics 91(3-4), 437–450.
Meyer, B. D. (1998). Do the Poor Move to Receive Higher Welfare Benefits? JCPR Working
Papers 58.
17
Nannestad, P. (2007). Immigration and Welfare States: A Survey of 15 Years of Research.
European Journal of Political Economy 23(2), 512–532.
Pedersen, P. J., M. Pytlikova, and N. Smith (2008). Selection and Network Effects: Migration
Flows into OECD Countries 1990–2000. European Economic Review 52(7), 1160–1186.
Pellizzari, M. (2012). The Use of Welfare by Migrants in Italy. International Journal of Manpower,
Special Issue on “Migration, the Welfare State, and European Labor Markets” (forth-
coming).
Razin, A., E. Sadka, and B. Suwankiri (2011). Migration and the Welfare State: PoliticalEconomy
Policy Formation. Cambridge, US: MIT Press.
Razin, A. and J. Wahba (2011). Welfare Magnet Hypothesis, Fiscal Burden and Immigration
Skill Selectivity. NBER Working Paper 17515.
Riphahn, R. T., M. Sander, and C. Wunder (2010). The Welfare Use of Immigrants and Natives
in Germany: The Case of Turkish Immigrants. LASER Discussion Paper 44.
Southwick, L. (1981). Public Welfare Programs and Recipient Migration. Growth and
Change 12(4), 22–32.
Zimmermann, K. F., M. Kahanec, A. Barrett, C. Giulietti, B. Maˆıtre, and M. Guzi (2012).
Study on Active Inclusion of Immigrants. IZA Research Report 43.
18