Migrant deprivation,conditionality of legal status and the welfare state

Abstract

This article adds to our theoretical understanding of the determination of third-country national (TCN) migrant deprivation and poverty in western Europe. The stratifying effects of different types of legal status on migrant outcomes have been established in previous research. The conditionality that states attach to securing different types of legal status has heretofore been overlooked as an important explanatory factor, however. A measure of the conditionality attached to attaining the key social rights–granting status of long-term residency (LTR) is operationalized using cross-national policy data. Building on existing theory, we hypothesize that the negative impact of welfare generosity on TCN material deprivation is moderated by a state’s level of LTR conditionality, such that deprivation will be greatest where conditionality is high and generosity is low. This hypothesis is tested using large-scale European microdata in the context of multilevel modelling. The empirical results are consistent with the central hypothesis. These findings have implications for policymakers and for extant accounts of migrant welfare, the welfare state and the factors implicated in the determination of poverty and deprivation in Europe.

Keywords

Comparative welfare states immigrants material deprivation migrant welfare multilevel modelling poverty

Introduction

Why are migrants poor? For modern European welfare states, migration is not only an established demographic fact, but also a contentious political issue, a challenge for societal integration and an ever-present policy concern. Research has shown that migrant households experience poverty with much greater frequency than native households (Sainsbury, 2012). This is an issue of basic social justice but also has broader implications. The exclusion of those who will always constitute a permanent minority may well foment social discord, where migrants adopt ‘oppositional stances’ to their host countries and natives resent perceived welfare abuse and dependency, with consequences for support for the welfare state itself and for social cohesion more generally (Banting and Kymlicka, 2006; Crepaz, 2008; Nannestad, 2007; Portes, 1997).

In Europe, third-country nationals (TCNs) face specific challenges regarding poverty. Quite apart from the deficits of language skills, cultural capital and social networks that such migrants experience next to intra-European migrants, TCNs suffer an added legal disadvantage insofar as they are often denied access to the protections of the welfare state by virtue of their legal status. While intra-European Union (EU) migrants have recourse to supranational legal directives at union level guaranteeing their access to the social protections of their host countries, TCNs do not.

Literature

Classic welfare regime theory in the vein of Esping-Andersen (1990) fails to account satisfactorily for the welfare outcomes of migrants, and it remains puzzling as to why exactly such stark differentials in, say, the poverty-alleviating capacities of welfare states should obtain even where states share similarities and belong to the same welfare regime. Research by Morissens and Sainsbury (see Sainsbury, 2012) on the income poverty experienced by TCNs showed little patterning by welfare regime type for post-transfer poverty rates; within the generous Scandinavian regime, for example, the poverty rate for Finland was twice that for Denmark (Sainsbury, 2012).

Similarly, a study of Denmark and Sweden found that the difference in poverty risk between immigrants and natives widened through the 1980s and 1990s but was more pronounced in Denmark (Blume et al., 2005). In part, this was due to the different types of migrant each country was receiving, and it was also seen that for certain migrants poverty decreased with years since migration (Blume et al., 2005). Other migrant-specific factors, such as language-deficits, social network deficits, human capital deficits, discrimination, and ethnicity, have also been seen to bear on related migrant outcomes like poverty and occupational attainment (Chiswick et al., 2005; Drinkwater and Clark, 2008; Platt, 2005, 2011; Sainsbury, 2012).

The major migrant-specific factor of interest in this article concerns the basis on which migrants enjoy rights and the role of legal status in that regard. Soysal (2012) has argued that rights expansion for migrants has occurred through supranational institutions, beyond the reach of the nation state, and that migrants now enjoy the rights of citizenship without necessarily possessing that legal status. Others have perceived a general convergence in the modes by which states ascribe citizenship status to non-nationals in Europe (Klusmeyer and Aleinikoff, 2000; Bloemraad, 2000; Castles, 2002; Joppke, 2007; Joppke and Morawska, 2003), and some authors have long maintained that migrant rights, including social rights, accrue on the basis of denizenship or mere residence-in-the-country (Brubaker, 1989; Cohen, 1987; Feldblum, 2000; Hammar, 1985). With such convergence, we might expect to see less diversity across states in terms of migrant poverty outcomes.

Sainsbury (2006, 2012) finds this account too simplistic, however, and rejects the idea that residence itself is sufficient to secure substantive, as opposed to merely legal, rights. Her work has highlighted the importance of immigration regimes for migrant social rights and has shown how different types of migrant entry category have stratifying effects on migrant social outcomes. Soysal also recognizes the contingency of migrant rights, noting that their expansion in the EU has been accompanied by the development of an ‘integration programme’, which, while supranational in nature, is practiced in differentiated ways across nation states (Soysal, 2012: 11). States demand integration by imposing formal requirements (e.g. tests of language and cultural understanding) and then tie this to the attainment of different types of legal status (Soysal, 2012).

Secure enjoyment of social rights and welfare entitlements in Europe accrues to those holding long-term residency (LTR) status, as numerous authors have observed (Bauböck, 2006; Brubaker, 1989; Dörr and Faist, 1997; Hammar, 1990) and as recent international data on migrants and migration policy collected by the Mipex project demonstrate (Niessen et al., 2007). LTR status also offers greater security than antecedent statuses, and so its acquisition can be expected to play a substantial role in securing TCN welfare. Across Europe, the empirical reality of the frameworks which function to formalize and categorize the migrant–state relationship is one of differentiated legal statuses varying from country to country in the nature and expansiveness of the rights they afford, as Mipex also makes clear (Niessen et al., 2007).

What has been neglected to date in studies of migrant poverty is the processual nature of migrant engagement with the regulatory frameworks which determine entry into different types of legal status. It is not the case that legal status is a static characteristic of migrant lives. The incentives offered by acquiring a more secure or expansive type of status can naturally be expected to have an impact on migrant actions and evaluations. The acquisition of progressively more secure legal status is not simply a matter of box-checking or pro forma approval on the part of the state and citizenship or LTR status must now be ‘earned’ (Soysal, 2012).

The acquisition of such status is a contingent process, conditional on fulfilling a set of specified requirements which themselves vary on a state-by-state basis in the stringency of the demands they make on migrants. What role does this ‘conditionality’ of legal status play in the determination of TCN migrant poverty outcomes? This is the research question with which this article engages. The key aim and academic contribution of the article is to expand our theoretical understanding of TCN poverty by explicating the hypothesized moderating effect that conditionality is expected to play on the relationship of state welfare effort to poverty.

Theory

On the determination of migrant poverty, the theoretical understanding adopted here shares a conceptual affinity with recent work conducted in the field of migration and welfare in Europe, on social protection and the ‘challenges of integration’ (Carmel et al., 2011). This work takes as one of its central themes the ‘interaction of migration, migration policies and social protection in Europe’ (Carmel and Cerami, 2011: 1; original emphasis).

This focus on the interaction of state structural and policy features is precisely the approach taken here. What is emphatically not hypothesized here is an abstracted, mono-causal framework, wherein conditionality plays the primary role in the determination of poverty. The inherent complexity of the institutional and policy superstructure within which migrants are embedded on arrival in the host country is explicitly acknowledged. What are the elements of this complex superstructure?

The institutional dimension of migrant integration has been conceived in terms of three policy domains: citizenship and immigration policies, social welfare policies, and labour market policies (Papadopoulos, 2011: 37). The aims of the research question addressed here, however, demand the articulation of an augmented theoretical viewpoint. Poverty, in this view, is taken to be one of several end-points of a process unique to the experience of TCN migrants.

An explicit assumption here is that migrants’ rights are clearly separate from those of prior residents, for example, native inhabitants, in welfare systems and that among migrants there are ‘clear distinctions made between legal and policy categories or status of migrant and the rights that attach to these categories’ (Carmel and Cerami, 2011: 6). Legal status is conceptualized as a hierarchically delimited continuum with TCNs at different points on that continuum being subject to varying levels of security and having recourse to different sets of rights. The constraints imposed on transitioning from a less secure or expansive point to a more secure or expansive point on that continuum have been characterized here as conditionalities, see Figure 1.

Figure 1.

Conceptual hierarchy of legal status, conditionality and security for TCNs.

The conditionality attached to securing a more favourable legal status is held to be partially determinative of poverty for TCN migrants as follows: high conditionality attached to long-term residency status will act to prevent full access for migrants to the welfare system while also entailing that migrants will subsist in more insecure life and work situations for longer periods of time. Thus, conditionality is centrally located in the nexus of TCN poverty determination, playing a key mediating role for migrants vis-a-vis the welfare system. In other words, the relationship of the poverty-alleviating capacity of the welfare state to the phenomenon of TCN migrant poverty is interactive with a state’s level of conditionality. Intuitively, we expect to see the highest levels of deprivation when policy parameters are at their most disadvantageous for TCNs, where conditionality is high and welfare effort is low.

Alongside the policy measures attached to legal status imposed at the national level, we also expect other national-level characteristics to have an impact on poverty outcomes and to facilitate explanation of cross-national variation in these outcomes. Such characteristics include the prevalence of anti-migrant attitudes (DeVoretz and Pivnenko, 2005) and the general economic situation including unemployment (Moller et al., 2003). These are not the only macro-level factors likely to be relevant but, due to space constraints, we leave it to future research to explore these other factors. In addition to the macro-level, it will be important to control for micro-level determinants of poverty, including socio-demographic characteristics such as age, sex, education level, marital status, employment status and family structure.

It should be stressed that this conception is concerned specifically with the role of conditionality of legal status and not with legal status itself. Lacking any data at the micro-level sufficient to disentangle long-term resident migrants from migrants resident on a more temporary basis, and to disentangle those from migrants who are naturalized citizens, it would simply not be possible to assess the direct effects of holding different legal statuses on poverty. The available microdata to be used here do not allow for such granularity of identification.

To this, it could be objected that adequate, aggregated data at the macro-level recording the proportions of TCNs comprising temporary, LTR and other categories could be used as macro-level indicators in the prediction of cross-national migrant poverty rates. This is, however, inadequate on two counts: (1) ‘Legal status’ is not a policy, and the aim here is to assess the role of state policies in the determination of migrant poverty; (2) a post facto indicator of the proportions of those holding specific legal statuses might well account for variation in poverty rates, but would do so without any reference to process, and it is this under-theorized aspect of TCN poverty determination that we wish to explicate. On the basis of our theoretical understanding of how and why the conditionality attached to long-term residency status seems likely to matter, we derive a central hypothesis:

Hypothesis: The negative impact of welfare generosity on TCN material deprivation and poverty is moderated by a state’s level of LTR conditionality such that deprivation will be greatest where conditionality is high and generosity is low.

Methodology

Data

In order to test this hypothesis, we utilize cross-national microdata in the form of the EU Statistics on Income and Living Conditions (EU-SILC) procured from Eurostat under licence, pooled for the years 2006 and 2007 (EU-SILC, 2006, 2007). The data are analysed by means of multilevel modelling, allowing for the testing of contextual level (level 2) interactions while controlling for individual-level (level 1) determinants of poverty. A level 2 measure of LTR conditionality is constructed for the developed countries of western Europe and its effects estimated using microdata from 14 countries.¹ EU-SILC data are collected irrespective of language, nationality or legal residence status (under European Commission Regulation No 1982/2003), and so we can be sure that the migrant population of interest is represented in the data. More information on the dataset can be found in the SILC supporting documentation (Eurostat, 2006).

Two years of cross-sectional data (2006, 2007) are merged, allowing us to boost sample numbers at level 1 and also to boost sample size at level 2. This leaves us with microdata from 14 countries (Germany and Switzerland unavailable), spread over 28 country-years and a total effective sample size of about 24,159 TCNs. The dataset is unbalanced insofar as some country-years contain more individual-level observations than others, averaging about 850 per country-year cluster, with a minimum of 150 observations and a maximum of 1745.²

We focus on the ‘rich’ developed countries of western Europe which we expect will have been desirable in similar and comparable ways to TCNs making relocation decisions. The logic at work here – in a broad sense – takes its cue from the ‘most similar systems’ comparative research design (Dogan and Pelassy, 1984; Peters, 1998). The point of this approach to comparative research is to use the research design to conceptually ‘hold constant’ as many possible sources of extraneous variance.

TCN migrant cases were identified from the data as follows. ‘Country of birth’ is defined as the country of residence of the mother at the time of birth with the variable taking three values (Eurostat, 2006: 156): same country as country of residence (LOC), any EU country except country of residence (EU25) and any other country (OTH). TCNs were defined as all those from ‘any other country’ on this variable. The regression models are constrained to TCNs only.

The dependent variable is a measure of material deprivation which is estimated as a function of a constant plus a vector of level 2 determinants, a vector of level 1 controls, a random intercept (the level 2 unit-specific deviation from the overall regression line, specific to each level 2 unit and constant across all individuals within that unit) and an error term specific to each individual within each level 2 unit.

Dependent variable

We operationalize poverty as a 7-point index of material deprivation following Whelan and Maître’s (2007) approach, with one amendment. Material deprivation is chosen over an alternative measure of, say, income poverty. It has been well documented that income is not always, or even often, a reliable predictor of deprivation: Income data at the lower and upper ends of the distribution may be unreliable; certain types of income, for example, from self-employment or investments, are more likely to be under-reported; households may have savings or lines of credit that allow for consumption beyond reported income just as they may have debt which requires servicing; reported income in a given year may be atypical due to unexpected factors such as illness or temporary unemployment (Kenworthy, 2007; Nolan and Whelan, 1996).

Other migrant-specific factors may reduce disposable household income, such as obligations to non-resident family and the remittance of monies to the source country.³ By contrast, material deprivation should more accurately gauge long-run income among those at the lower end of the distribution, a view supported by a 2006 panel data study (Boarini et al., 2006, cited in Kenworthy, 2007) which found that 70 percent of those identified as materially deprived at any given time point were persistently in that state over the 4-year period of the study. This is important: this study is cross-sectional in nature, and reliance on income measures will be undesirable where such measures are subject to temporary fluctuations.

Also, a focus on deprivation allows for an explanation of effects of variables in terms of their contribution to ‘more or less’ poverty or deprivation, something not possible with a binary ‘poor/not poor’ outcome variable. We are not interested in head-counting the poor but rather in assessing how structural parameters and the dynamics created by legal status, and the conditions attached to legal status, interact to produce poverty outcomes that vary in intensity for TCNs across Europe. Nevertheless, as a precautionary check on the results, we also estimate models using a binary measure of the income poverty threshold as the dependent variable.

Whelan and Maître (2007) formed an 8-point index using indicators from the ‘economic strain’ in the set of EU-SILC’s deprivation items, where this strain deals with deprivation relating to the inability to afford (or experiencing the enforced lack of) various basic needs. The specific eight items on which deprivation was measured were as follows: regular healthy meals, having a car, telephone, PC, or annual 1-week holiday, the inability to cope with unexpected expenses, inability to adequately warm the home, and ‘arrears relating to mortgage/rent or hire-purchase’. The indicators are mostly binary outcomes, and the index was formed by assigning a score of 1 to those people recording deprivation for that indicator and then summing the responses to form an index where higher scores indicate greater deprivation.

The methodological advantage of following this approach is that these authors have already analysed and assessed the reliability, consistency and loadings of the index generated by utilizing these items, finding that the items load on a single factor in a ‘homogeneous manner’ with a satisfactory Cronbach’s alpha score of 0.76 (Whelan and Maître, 2007: 157). The amendment to this procedure made here entailed dropping one of the items, the ‘arrears relating to mortgage, rent or hire-purchase’ item, due to severe missing data problems for this variable.⁴ This 7-point index of material deprivation has a mean of 1.7 and standard deviation of 1.6.

There is some academic disagreement over what deprivation items should be considered ‘necessities’ (Kenworthy, 2007), for example, taking an annual holiday may not be equivalent to being able to afford regular healthy meals. To allow for this, we conduct a sensitivity analysis using alternative operationalizations of the dependent variable – namely individual binary indicators for the capacity to warm the family home, to afford regular healthy meals, and to deal with unexpected expenses – as well as an additive 3-point index of these essential items.

Independent variables

To operationalize our measure of LTR conditionality, we build on the data collected by the Mipex project for European migration policies in 2006 (Niessen et al., 2007). In the project’s own words, ‘Mipex measures policies to integrate migrants in 25 EU Member States and three non-EU countries. It uses over 140 policy indicators to create a rich, multidimensional picture of migrants’ opportunities to participate in European societies. Mipex covers six policy areas which shape a migrant’s journey to full citizenship: labour market access, family reunion, long-term residence, political participation, access to nationality and anti-discrimination (Niessen et al., 2007: x).’

Drawing on the policy strand associated with the conditions attached to securing LTR status for 16 countries of western Europe, it was possible to construct an index of conditionality utilizing the country-by-country data which was made available for download on the Mipex project’s website. There were seven component policy measures within this strand which clearly met the criteria of ‘imposing a condition’ on TCN migrants.⁵ In the original Mipex dataset, each measure was coded for each country as 1, 2 or 3, where 1 indicated the most exacting level of requirement and 3 the least. The polarities were reversed for present purposes such that in the formation of an additive index, high values would correspond to high levels of conditionality. The initial set of seven policy measures, before reduction using statistical techniques, are represented in Table 1.

Table 1.

Policy indicators for conditionality of long-term resident status.

	1	2	3
Required time of habitual residence	≤5 years	>5 ≤ 8 years	>8 years
Required time in legal employment	≤3 years	>3 ≤ 5 years	>5 years
Economic resources requirement:	None	A	A and/or B
A. Employment-related criteria
B. Stable and sufficient means for applicant and dependents
Insurance requirement	None	Simple sickness insurance required	Other type of insurance (all risks, etc.)
Length of application procedure	≤ 6 months	>6 ≤ 9 months	> 9 months or no regulation on max length
Costs of application and/or issue of permit or renewal	None	Administrative fee as charged for issue of identity card	Any higher costs or prerequisites at high indirect costs (e.g. medical records)
Integration measures (overall average)	Average of indicators below	Average of indicators below	Average of indicators below
Integration measures	None	Other integration measures	Passing of test
Imposition of integration course	None	Voluntary	Conditional for status
Language assessment	None	Simple, oral, multiple choice	Written and/or high-level language test
Content of integration assessment	None	With social aspects (knowledge of legal/political system, citizenship rights, basic norms/values)	With cultural aspects (culture, customs, traditions)

Source: Review of indicators for the 2006 Migrant Integration Policy Index: Excel version; available from Mipex website (http://www.mipex.eu/).

Multiple correspondence analysis (MCA) was used to assess the dimensionality of the measure to be extracted from these data. MCA is used here as the non-linear cognate of principal components analysis (PCA), where PCA is a standard tool in the social sciences used to reduce higher-dimensional vector spaces (of indicators) to lower-dimensional constructs which may then be used as indexes (Lewis-Beck, 1994). MCA is applied to categorical/ordinal data that produces results analogous to PCA by analysing the correspondences between rows and columns in a data matrix (Le Roux and Rouanet, 2004: 180). The matrix in this instance is a 16 × 7 matrix, for 16 countries over seven policy measures, see Table 2. Whereas in standard PCA, the quality of a solution is affected by sample size, in MCA, sample size is not relevant to the construction of the map charting the relationships between variables (Greenacre and Blasius, 2006: 14).

Table 2.

Conditionality values for matrix of policy indicators by country.

Country	HabitualRes	TimeEmpl	Integration	EcResource	Insurance	AppLength	Cost
Austria	1	1	3	3	3	1	3
Belgium	1	1	1	1	1	2	2
Denmark	2	1	2	2	1	1	1
Finland	1	2	1	3	1	1	2
France	1	1	2	3	2	3	3
Germany	1	2	3	3	1	3	3
Greece	1	2	3	3	2	2	3
Ireland	2	2	1	1	1	1	2
Italy	1	1	1	3	2	1	2
Netherlands	1	2	3	3	2	1	3
Norway	1	1	2	3	1	2	2
Portugal	2	1	1	3	1	1	2
Spain	1	1	1	1	1	1	2
Sweden	1	1	1	3	1	1	2
Switzerland	3	1	2	3	1	1	3
United Kingdom	3	2	1	1	1	2	2

The analysis indicated that the ‘time employed’ indicator and the ‘length of application’ indicator did not map to the underlying construct in a satisfactory unidimensional manner, and so these were excluded from the final measure. To save space, the results of the MCA are not shown here (see Supplementary Table A1). Consulting Cronbach’s alpha, a standard means of computing inter-item correlations and a test statistic for the internal consistency, that is, reliability, of an index, showed that the ‘length of habitual residence’ indicator was relatively poorly correlated with the rest of the items and with the overall index. Lack of variation in the indicator most likely accounts for this poor correlation. Most countries (10 of the 14 for which we have microdata) take the same value on this indicator, requiring 5 years or fewer habitual residence, and only one country (United Kingdom) imposes the most stringent requirement recorded in these data, of more than eight years required residence. Excluding length of residence from the final measure gave an alpha score of 0.80, where values of alpha ranging from 0.72 to 0.88 have been cited as indicating acceptable to high reliability (John and Benet-Martínez, 2000). Summing the remaining four indicators produced an 8-point additive index of conditionality with minimum values pegged to zero; see Table 3 for the final variable with each country’s corresponding score along with summary data for the variable.

Table 3.

Conditionality and welfare effort scores by country.

Country	LTR conditionality	Welfare effort
Austria	8	6.34
Belgium	1	10.74
Denmark	2	9.95
Finland	3	6.50
France	6	6.05
Germany	6	7.73
Greece	7	5.23
Ireland	1	6.4
Italy	4	3.86
Netherlands	7	10.87
Norway	4	11.79
Portugal	3	10.70
Spain	1	7.84
Sweden	3	10.03
Switzerland	5	9.70
United Kingdom	1	5.14
Mean	3.88	8.05
Standard deviation	2.40	2.49

To operationalize welfare effort, we build on the approach taken by Scruggs and Allan (2006a, 2006b) in their replications and extensions of Esping-Andersen’s (1990) decommodification index. This produced a measure capturing the relative differences in unemployment benefit generosity (in 2000) for a set of Organisation for Economic Co-operation and Development (OECD) countries on the basis of empirical data in five areas: replacement rates, the duration of benefits, the qualifying conditions that apply to receipt of those benefits, waiting days that may be imposed before claimants receive benefits, and the percentage of the population covered by the benefits in question.

A focus on welfare supports directly associated with loss of employment seems appropriate given that migrant-specific challenges has contributed to migrants’ higher risk of unemployment (Chiswick et al., 2005; Drinkwater and Clark, 2008; OECD, 2013). Looking to other aspects of social protection with a strong contributory element, such as pensions, would arguably be inappropriate where migrants are the focus of the analysis. Migrants often have incomplete work histories or tenuous labour market attachments which militate against the full enjoyment of long-run contributory benefits (Dörr and Faist, 1997). In some countries, certain categories of TCN face expulsion if they become unemployed, reliant on welfare payments, or after a defined period on such payments (Niessen et al., 2007).

Measures capturing overall levels of social expenditure, which include things like health spending, will arguably be less relevant to material deprivation; we do not necessarily expect a correlation between state healthcare spending and the ability to heat one’s home. As such, we expect employment-related cash benefits to matter more for migrant populations. As a precautionary check on the main findings, however, we will later estimate models utilizing alternative measures of welfare and social spending drawn from OECD data.

Scruggs and Allan do not address themselves to all of the countries in this study, specifically omitting consideration of Greece, Portugal and Spain. Applying these authors’ methodology using more recent data, however, it was possible to reconstruct the benefit generosity index to include the three omitted countries. For purposes of consistency and comparability, the Scruggs–Allan approach was followed as closely as possible, and this approach has been outlined elsewhere (Scruggs and Allan, 2006a: 891–92). Data were collected from several sources: the OECD (2004, 2009), the International Social Security Association (ISSA, 2008) and the most recently updated data from the Scruggs dataset on welfare state entitlements made available on that author’s website⁶ (Scruggs, 2004). Table 3 records each country’s score on this measure, along with summary data.

In the models of TCN material deprivation, we include individual-level determinants covering a standard set of socio-demographic indicators held to be generally relevant in poverty research including: age, sex, education, household type, marital status and employment status (Nolan and Whelan, 1996; Cappellari and Jenkins, 2002; Reinstadler and Ray, 2010). For migrants specifically, research has shown that those who have acquired naturalized citizenship status also have lower poverty rates (Sainsbury, 2012), so we also control for this.

The unemployment rate is likely to be an important macro-determinant of poverty. As individuals experience a loss of income due to unemployment, and because real wages often decline during economic downturns, the theoretical expectation is that higher rates of unemployment will be correlated with higher levels of poverty (Moller et al., 2003). The variable used in the analysis is taken from OECD data and captures the unemployment rate of the civilian population overall for the year preceding collection of our microdata, that is, 2005 (OECD, 2007: 36–7).

High levels of anti-migrant attitudes are likely to result in greater discrimination against migrants, leading to greater social distance in a way that impacts hiring decisions and thus results in greater poverty (DeVoretz and Pivnenko, 2005). To gauge the extent to which native populations in our countries of interest express negative sentiments towards migrant populations, we have recourse to directly collected first-hand survey data in the form of the European Social Survey (ESS) for 2006, to correspond with the time period for the SILC microdata.⁷

The question utilized for the present purposes was as follows: ‘Is [country] made a worse or a better place to live by people coming to live here from other countries?’, with respondents asked to respond on a scale from 0 to 10 where ‘0’ corresponds to ‘Worse place to live’ and ‘10’ corresponds to ‘Better place to live’. Recoding this scale into a binary variable, such that responses 0 through 4 were assigned a value of ‘1’ and all other responses assigned a value of ‘0’, it was possible to determine the proportion of total respondents in each country subscribing to these anti-migrant sentiments (i.e. taking a value of ‘1’ on the binary recoded variable). These proportions were taken as the values for the final level 2 indicator.

Multilevel modelling

Given the unbalanced nature of the data, maximum likelihood (ML) and restricted maximum likelihood (REML) are the ‘preferred methods of estimation’ over other alternatives (Marchenko, 2006). It has been noted that a sufficient number of level 2 units for random effects modelling is ‘typically more than 10 or 20’ (Rabe-Hesketh and Skrondal, 2008: 62). Maas and Hox have shown in a simulation study that with 30 level 2 units, using REML estimation, regression coefficients and standard errors were estimated reliably (Maas and Hox, 2005).

The choice was made to model the random effects here as country-years so as to increase our level 2 ‘sample size’ from 14 to 28, that is, 14 countries over 2 years. Utilizing country-year clusters and individual responses is by now an established practice in multilevel research (e.g. Arceneaux and Nickerson, 2009; Jensen and Lindstadt, 2005). As a check on the results, however, the random effects were also modelled as 14 country units. The analyses conducted are cross-sectional and, as the respondents are independent in each year of the survey used, the data does not constitute a panel. Weights were not applied to the data. While EU-SILC does include a set of weights, these have been constructed with reference to the full population, that is, on the basis of both nationals and non-nationals alike. Application of these weights to a migrant-specific analysis would not be justified and could well make the sample less representative (Barrett and McCarthy, 2007: 807).

Results

Estimating the null model (not shown) returns a significant random intercept, the standard deviation of which is 0.522. The constant in this model is 1.645, while the standard deviation of the residual is 1.515. Another model (not shown) using only level 1 predictors returns a random intercept with standard deviation of 0.426, while the standard deviation of the residual is 1.428. It is important to control for individual-level factors in the context of multilevel modelling so as to be sure that any inter-group differences (between the level 2 units) detected in the outcome variable are not actually simply the result of differences in the composition of migrant groups from country to country (i.e. in the characteristics of the individuals of whom the groups consist), otherwise known as compositional effects (Snijders and Bosker, 1999).

The reduction in the random intercept from baseline value 0.522 to 0.426 with the addition of level 1 predictors implies a level 2 R² of 33 percent.⁸ This indicates that one-third of the variance in material deprivation across level 2 units is accounted for by characteristics at the individual level. For space reasons, models using only level 2 indicators are not shown here, but they indicate the following: no effect for LTR conditionality in isolation, lower levels of material deprivation are significantly predicted (p < 0.10) with the addition of the welfare effort indicator, and interacting benefit generosity and LTR conditionality returns a highly significant interaction effect.

The first model in Table 4 presents the full interactive model, controlling for the set of determinants at the individual level. The random intercept for this full model has standard deviation of 0.317, implying a level 2 R² of 63 percent. The central level 2 interaction of interest is significant even controlling for level 1 predictors. Substantively, the results presented in this table tell us that – even when controlling for the individual-level determinants of poverty – the generosity of the welfare benefits system in interaction with the conditionality attached to receiving those benefits for TCN migrants works to determine migrant deprivation in significantly different ways across countries. Benefit generosity plays an independent predictive role with regard to deprivation, as we would expect, though these results indicate that this effect is moderated by the level of conditionality attached to attaining LTR status, consistent with the overarching hypothesis. The positive main effect of welfare effort in this and other models is interpreted as the effect when LTR conditionality is equal to zero, though the effect is negative when conditionality is at mean or higher levels.

Table 4.

Central interaction of level 2 variables with level 1 and level 2 controls.

	1	2	3	14 units
L2: LTR conditionality	0.404***	0.276**	0.323***	0.405**
	(0.093)	(0.101)	(0.065)	(0.136)
L2: Welfare effort	0.152**	0.102*	0.140***	0.152**
	(0.052)	(0.052)	(0.036)	(0.076)
L2: LTR*Welfare Effort	−0.053***	−0.036**	−0.046***	−0.053**
	(0.012)	(0.013)	(0.008)	(0.018)
Unemployment rate (2005)		0.081**
Unemployment rate (2005)		(0.034)
Anti-migrant attitudes (2002)			0.021***
Anti-migrant attitudes (2002)			(0.004)
Naturalized (0, 1)	−0.520***	−0.521***	−0.520***	−0.520***
	(0.020)	(0.020)	(0.020)	(0.020)
Age (low)	−0.052**	−0.052**	−0.052**	−0.050*
	(0.026)	(0.026)	(0.026)	(0.026)
Age (high) [ref: mid]	0.113***	0.112***	0.112***	0.111***
	(0.023)	(0.023)	(0.023)	(0.023)
Sex (F = 1)	−0.039**	−0.039**	−0.039**	−0.040**
	(0.019)	(0.019)	(0.019)	(0.019)
Primary	0.927***	0.925***	0.926***	0.917***
	(0.033)	(0.033)	(0.033)	(0.033)
Lower secondary	0.539***	0.539***	0.537***	0.529***
	(0.028)	(0.028)	(0.028)	(0.028)
Upper secondary [ref: tertiary]	0.182***	0.182***	0.181***	0.173***
	(0.024)	(0.024)	(0.024)	(0.024)
Married (0, 1)	−0.317***	−0.317***	−0.317***	−0.316***
	(0.022)	(0.022)	(0.022)	(0.022)
Unemployed (0, 1)	0.731***	0.731***	0.733***	0.731***
	(0.034)	(0.034)	(0.034)	(0.034)
Single parent (0, 1)	0.786***	0.786***	0.787***	0.787***
	(0.049)	(0.049)	(0.049)	(0.049)
Household size	−0.034***	−0.034***	−0.034***	−0.034***
	(0.006)	(0.006)	(0.006)	(0.006)
Constant	0.648	0.477	0.076	0.653
	(0.425)	(0.396)	(0.311)	(0.621)
Random: SD (intercept)	0.317***	0.289***	0.213***	0.331***
	(0.048)	(0.045)	(0.034)	(0.075)
SD (residual)	1.428***	1.428***	1.428***	1.429***
	(0.007)	(0.007)	(0.007)	(0.007)
N	23573	23573	23573	23573
F-test	0	0	0	0
AIC	83886.943	83888.596	83879.012	83891.751
Chi-square	3018	3032	3100	2975
Log-likelihood	−41926	−41926	−41922	−41929

AIC: Akaike Information Criterion; LTR: long-term residency; SD: standard deviation.

Models 1, 2 and 3 use 28 level 2 units.

p < 0.10; **p < 0.05; ***p < 0.001.

The next two models control for other important contextual determinants of deprivation, alongside the central interaction of interest. Model (2) controls for a country’s unemployment rate. As expected, increasing unemployment positively and significantly predicts higher levels of material deprivation. Model (3) finds a highly significant effect for anti-migrant attitudes impacting on migrant deprivation, in the expected direction. The final model (4) is a re-estimation of model 1 using 14 country groupings as the level 2 unit. As can be seen, this more cautious approach yields significant results in line with the central hypothesis.

The individual-level effects are all significant at 5 percent and run in the expected direction: those with lower levels of education relative to those with third-level education are all predicted to experience higher levels of deprivation; the unemployed and single-parent families are also predicted to have higher levels of deprivation, relative to all other employment and family structure statuses, respectively; those who are older/elderly experience more deprivation than those of prime working age; married households experience less poverty than those of any other marital status; larger households endure less deprivation, likely due to having more economically active members. Naturalized TCN migrants experience less deprivation than non-naturalized migrants, even controlling for other factors.

We can graph the interaction effect for a clearer understanding of the dynamic at work. As can be seen in Figure 2, as welfare generosity increases to its highest levels, poverty is predicted to converge to the same point regardless of the level of conditionality.⁹ This is as expected: more welfare means less poverty. More precisely, at mean conditionality (dashed line), the predicted level of deprivation decreases as welfare effort increases, as expected.

Figure 2.

Effect of welfare effort on material deprivation, varying levels of LTR conditionality.

Two points below the mean (line A), the regression line is practically flat, indicating no significant effect, or that at low levels of conditionality, predicted poverty is low regardless of level of welfare benefits. At maximum conditionality (D), the intercept for poverty is very high at low levels of welfare effort, and predicted poverty decreases as welfare levels increase. These results indicate a relationship between conditionality and welfare levels in line with theoretical expectations. They indicate that higher conditionality affects TCNs most detrimentally in welfare systems with low levels of support and protection.

As a confirmatory check, we also utilized an alternative welfare effort indicator. Operationalizing welfare effort as the amount of social spending assigned to income support, using OECD data collected in 2006 (OECD, 2011), returned a significant interaction in line with the findings above (see Supplementary Table A2). Other potential indicators, including a country’s overall expenditure on social spending or on public spending, are not significant in these alternative specifications, most likely for the reasons set out earlier concerning the importance of employment-related benefits to migrants and the decreased relevance of social protections requiring long contribution histories.

A sensitivity analysis using different operationalizations of the dependent variable returned results in line with the findings just discussed (see Supplementary Table A3).¹⁰ Finally, models taking a binary income poverty threshold measure as the dependent variable returned results indicating a significant interaction between welfare effort and conditionality (see Supplementary Table A4).

Discussion

Comparative European welfare state research draws contrasts between the modalities of social protection in Europe, facilitating the explanation or prediction of variable life and work outcomes for the inhabitants of European welfare states. Yet, with continued disparities of wealth and living standards obtaining globally and with the secular transformation of human society such that transglobal communication and transportation are now possible, it is no longer the case that the inhabitants of Europe are all European inhabitants. One in 10 of those inhabitants, on average, will not have been born in the country where they live, while many will not have been born in Europe at all.

These TCNs occupy a unique and qualitatively different position with regard to the structural, institutional and legal frameworks which govern access to the protections of the welfare state. As such, the modalities of social protection in Europe are of little relevance to explicating TCN outcomes, and many extant academic accounts are inherently partial and insufficient to the task. A central aspect of this qualitative difference inheres in the role played by legal status in the regulation of TCN rights and entitlements.

In line with Sainsbury’s (2012) injunction to attend to the dynamics between welfare and immigration regimes, this research set out to explore the role of legal status in determining migrant outcomes. Cross-nationally, migrant poverty outcomes are at variance with the expectations of welfare regime theory (Esping-Andersen, 1990). The insight that the relationship of TCNs to the welfare state is qualitatively different for state nationals is itself a theoretical challenge to the existing regime-based accounts of welfare.

This article expands our theoretical understanding of the role of legal status in TCN welfare, beginning from the observations that the extension of rights-granting legal status is part of a process which TCNs must navigate and that the process itself is subject to substantial variation across states in terms of the conditionality imposed on attaining this status. The results of the empirical analysis are clearly consistent with the central hypothesis: conditionality of long-term resident status moderates the effect of welfare state generosity on migrant poverty outcomes such that deprivation is seen to be greatest where conditionality is high and generosity is low.

While migrants may enjoy many of the formal rights of citizenship without necessarily possessing that status (Soysal, 2012), their substantive as opposed to merely formal enjoyment is still contingent on the rules and regulatory practices that nation states adopt to manage membership (Feldblum, 2000; Sainsbury, 2012). The findings here regarding the role of conditionality serve to add depth to our understanding of the process by which states manage their members and the means by which states exercise their sovereignty so as to extend (or retract, or deny) rights, while simultaneously highlighting the consequences of different membership management strategies for migrant outcomes.

This means that conditionality must now be set alongside the host of other migrant-specific factors that we know to be implicated in the determination of migrant poverty. Claims regarding the importance of, for example, language and human capital skills, discrimination and ethnicity must now be interpreted with due regard given not only to the legal status of the migrants in question but also to the policy landscape governing legal status acquisition (Blume et al., 2005; Chiswick et al., 2005; Drinkwater and Clark, 2008; Platt, 2005, 2011).

Insofar as the conditionality attached to legal status is implicated in migrant welfare, then the rules and regulations around the extension of legal status, the tools of migration policy, can legitimately be seen as de facto tools of social policy. Such an analytical reappraisal may provide sufficient impetus for policymakers to seek a greater coherency of purpose regarding the macro-structural parameters delineating the welfare outcomes of migrant populations, where the alleviation of poverty is an active policy concern.

It should be stressed that conditionality itself is not hypothesized here as a primary determinant of poverty outcomes, rather, that it plays an important moderating role as regards the TCN–welfare state relationship. A limitation of this study concerns the constraint on available degrees of freedom at level 2, which places a limit on the inclusion of control variables. As further Mipex data become available, the pooling of multiple cross-sections would facilitate the inclusion of further controls or interactions while also opening up possibilities to explore conditionality and changes in migrant poverty over time. This was not possible in this study but could be fruitfully addressed in future research.

Fundamentally, this new and nuanced viewpoint highlights the contingent and conditional nature of TCN welfare in Europe while forcing a conceptual relocation of the locus of TCN migrant welfare out of the Work–Welfare nexus, within which native inhabitants find themselves, and into a more complex nexus delimited by the complementary interconnections of Work–Conditionality–Legal Status–Welfare.

Footnotes

Acknowledgements

The author would like to thank the anonymous reviewers, as well as Anthony McCashin and Philip Curry of Trinity College Dublin and Wim van Oorschot of Tilburg University for their helpful comments on earlier versions of this work.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Arceneaux

Nickerson

D.W.

(2009) ‘Modeling Certainty with Clustered Data: A Comparison of Methods’, Political Analysis 17(2): 177–90.

Banting

Kymlicka

(2006) ‘Immigration, Multiculturalism, and the Welfare State’, Ethics & International Affairs 20(3): 281–304.

Barrett

McCarthy

(2007) ‘Immigrants in a Booming Economy: Analyzing Their Earnings and Welfare Dependence’, Labour 21(4–5): 789–808.

Bauböck

(2006) Migration and Citizenship: Legal Status, Rights and Political Participation. Amsterdam: Amsterdam University Press.

Bloemraad

(2000) ‘Citizenship and Immigration: A Current Review’, Journal of International Migration and Integration 1: 9–37.

Blume

Gustafsson

Pedersen

P.J.

Verner

(2005) ‘A Tale of Two Countries: Poverty Among Immigrants in Denmark and Sweden Since 1984’, in Borjas

G.J.

Crisp

(eds) Poverty, International Migration and Asylum, pp. 317–40. Basingstoke: Palgrave Macmillan.

Boarini

Johansson

Mira d’Ercole

(2006) ‘Measures of Material Deprivation in OECD countries’, OECD Social, Employment, and Migration Working Paper 37. Available at: www.oecd.org/els/soc/37223552.pdf

Brubaker

(1989) ‘Membership without Citizenship: The Economic and Social Rights of Noncitizens’, in Brubaker

(ed.) Immigration and the Politics of Citizenship in Europe and North America, pp. 145–162. Lanham, MD: University Press of America.

Cappellari

Jenkins

S.P.

(2002) ‘Who Stays Poor? Who Becomes Poor? Evidence from the British household Panel Survey’, The Economic Journal 112(478): C60–7.

10.

Carmel

Cerami

(2011) ‘Governing Migration and Welfare: Institutions and Emotions in the Production of Differential Integration’, in Carmel

Cerami

Papadopoulos

(eds) Migration and Welfare in the ‘New’ Europe: Social Protection and the Challenges of Integration, pp. 1–22. Bristol: Policy Press.

11.

Carmel

Cerami

Papadopoulos

(eds) (2011) Migration and Welfare in the ‘New’ Europe: Social Protection and the Challenges of Integration. Bristol: Policy Press.

12.

Castles

(2002) ‘Migration and Community Formation under Conditions of Globalization’, International Migration Review 36(4): 1143–68.

13.

Chiswick

B.R.

Lee

Y.L.

Miller

P.W.

(2005) ‘A Longitudinal Analysis of Immigrant Occupational Mobility: A Test of the Immigrant Assimilation Hypothesis’, International Migration Review 39(2): 332–53.

14.

Cohen

(1987) The New Helots: Migrants in the International Division of Labour. Aldershot: Avebury.

15.

Crepaz

(2008) Trust Beyond Borders: Immigration, the Welfare State, and Identity in Modern Societies. Ann Arbor, MI: University of Michigan Press.

16.

DeVoretz

D.J.

Pivnenko

(2005) ‘The Economic Causes and Consequences of Canadian Citizenship’, Journal of International Migration and Integration 6(3): 435–68.

17.

Dogan

Pelassy

(1984) How to Compare Nations: Strategies in Comparative Politics. Chatham, NJ: Chatham House Publishers.

18.

Dörr

Faist

(1997) ‘Institutional Conditions for the Integration of Immigrants in Welfare States: A Comparison of the Literature on Germany, France, Great Britain, and the Netherlands’, European Journal of Political Research 31(4): 401–26.

19.

Drinkwater

Clark

(2008) ‘The Labour Market Performance of Recent Migrants’, Oxford Review of Economic Policy 24(3): 495–516.

20.

Esping-Andersen

(1990) Three Worlds of Welfare Capitalism. Oxford: Oxford University Press.

21.

Eurostat (2006) Description of SILC User Database Variables: Cross-Sectional and Longitudinal. European Commission, Version 2006.1, 1 March 08. Luxembourg: Eurostat.

22.

EU Statistics on Income and Living Conditions (EU-SILC) (2006) EU-SILC UDB 2006 – Version 4 of March 2010 (Cross-sectional microdata). Luxembourg: Eurostat.

23.

EU Statistics on Income and Living Conditions (EU-SILC) (2007) EU-SILC UDB 2007 – Version 4 of August 2010 (Cross-sectional microdata). Luxembourg: Eurostat.

24.

Feldblum

(2000) ‘Managing Membership: New trends in Citizenship and Nationality Policy’, in Klusmeyer

Aleinikoff

T.A.

(eds) From Migrants to Citizens: Membership in a Changing World, pp. 475–500. Washington, DC: Carnegie Endowment for International Peace.

25.

Greenacre

M.J.

Blasius

(2006) Multiple Correspondence Analysis and Related Methods. London: CRC Press.

26.

Hammar

(1985) ‘Dual Citizenship and Political Integration’, International Migration Review 19: 438–50.

27.

Hammar

(1990) Democracy and the Nation State: Aliens, Denizens and Citizens in a World of International Migration. Aldershot: Avebury.

28.

International Social Security Association (ISSA) (2008) Social Security Programs Throughout the World: Europe, 2008. Geneva: ISSA and Social Security Administration (US).

29.

Jensen

N.M.

Lindstadt

(2005) ‘Chasing the Firm or Rewarding the Partisans? Domestic Responses to International Tax Competition in the OECD, 1982–1998’, Paper presented at the American Political Science Association Annual Meeting, 1 September, Washington, DC.

30.

John

O.P.

Benet-Martínez

(2000) ‘Measurement: Reliability, Construct Validation, and Scale Construction’, in Reis

H.T.

Judd

C.M.

(eds) Handbook of Research Methods in Social and Personality Psychology, pp. 339–69. Cambridge: Cambridge University Press.

31.

Joppke

(2007) ‘Transformation of Immigrant Integration: Civic Integration and Antidiscrimination in the Netherlands, France, and Germany’, World Politics 59: 243–73.

32.

Joppke

Morawska

(eds) (2003) Toward Assimilation and Citizenship: Immigrants in Liberal Nation-States. Basingstoke: Palgrave Macmillan.

33.

Kenworthy

(2007) Measuring Poverty and Material Deprivation. Ottawa, Canada: Statistics Canada.

34.

Klusmeyer

D.B.

Aleinikoff

T.A.

(2000) From Migrants to Citizens: Membership in a Changing World. Washington, DC: Carnegie Endowment for International Peace.

35.

Le Roux

Rouanet

(2004) Geometric Data Analysis: From Correspondence Analysis to Structured Data Analysis. Dordrecht, The Netherlands: Springer.

36.

Lewis-Beck

M.S.

(1994) Factor Analysis and Related Techniques: International Handbook of Quantitative Applications in the Social Sciences, vol. 5. London: Sage/Toppan Publishing.

37.

Maas

C.J.

Hox

(2005) ‘Sufficient Sample Sizes for Multilevel Modeling’, Methodology: European Journal of Research Methods for the Behavioral and Social Sciences 1(3): 86–92.

38.

Marchenko

(2006) ‘Estimating Variance Components in Stata’, Stata Journal 6(1): 1–21.

39.

Moller

Huber

Stephens

J.D.

Bradley

Nielsen

(2003) ‘Determinants of Relative Poverty in Advanced Capitalist Democracies’, American Sociological Review 68(1): 22–51.

40.

Nannestad

(2007) ‘Immigration and Welfare States: A Survey of 15 Years of Research’, European Journal of Political Economy 23: 512–32.

41.

Niessen

Huddleston

Citron

(2007) MIPEX–Migrant Integration Policy Index. Brussels: British Council and Migration Policy Group (in cooperation with Andrew Geddes and Dirk Jacobs).

42.

Nolan

Whelan

C.T.

(1996) Resources, Deprivation, and Poverty. Oxford: Oxford University Press.

43.

Organisation for Economic Co-operation and Development (OECD) (2004) ‘Net Replacement Rates for Six Family Types, Initial Phase of Unemployment’, in Tax-Benefit Models. Paris: Organisation for Economic Cooperation and Development, available at http://www.oecd.org/dataoecd/25/28/34008439.xls.

44.

Organisation for Economic Co-operation and Development (OECD) (2007) OECD in Figures 2007. Paris: OECD.

45.

Organisation for Economic Co-operation and Development (OECD) (2009) Government at a Glance 2009. Paris: OECD.

46.

Organisation for Economic Co-operation and Development (OECD) (2011) Society at a Glance 2011: OECD Social Indicators. Paris: OECD.

47.

Organisation for Economic Co-operation and Development (OECD) (2013) International Migration Outlook 2013, vol. 2013. Paris: OECD.

48.

Papadopoulos

(2011) ‘Immigration and the Variety of Migrant Integration Regimes in the European Union’, in Carmel

Cerami

Papadopoulos

(eds) Migration and Welfare in the ‘New’ Europe: Social Protection and the Challenges of Integration, pp. 23–48. Bristol: Policy Press.

49.

Peters

B.G.

(1998) Comparative Politics: Theory and Methods. New York: NYU Press.

50.

Platt

(2005) Migration and Social Mobility: The Life Chances of Britain’s Minority Ethnic Communities. Bristol: The Policy Press.

51.

Platt

(2011) Inequality within Ethnic Groups. York: Joseph Rowntree Foundation.

52.

Portes

(1997) ‘Immigration Theory for a New Century: Some Problems and Opportunities’, International Migration Review 31: 799–825.

53.

Rabe-Hesketh

Skrondal

(2008) Multilevel and Longitudinal Modeling using Stata. College Station, TX: Stata Corp.

54.

Reinstadler

Ray

J.C.

(2010) Macro Determinants of Individual Income Poverty in 93 Regions of Europe. Luxembourg: EU Publications Office; Eurostat.

55.

Sainsbury

(2006) ‘Immigrants’ Social Rights in Comparative Perspective: Welfare Regimes, Forms in Immigration and Immigration Policy Regimes’, Journal of European Social Policy 16(3): 229–44.

56.

Sainsbury

(2012) Welfare States and Immigrant Rights: The Politics of Inclusion and Exclusion. Oxford: Oxford University Press.

57.

Scruggs

(2004) ‘Summary Dataset’, Welfare State Entitlements Data Set: A Comparative Institutional Analysis of Eighteen Welfare States, Version 1.2, available at http://sp.uconn.edu/~scruggs/cwed/sumdat11.xls.

58.

Scruggs

Allan

(2006a) ‘The Material Consequences of Welfare States: Benefit Generosity and Absolute Poverty in 16 OECD Countries’, Comparative Political Studies 39: 880–904.

59.

Scruggs

Allan

(2006b) ‘Welfare-state Decommodification in 18 OECD Countries: A Replication and Revision’, Journal of European Social Policy 16(1): 55–72.

60.

Snijders

T.A.

Bosker

(1999) Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modeling. London: Sage.

61.

Soysal

Y.N.

(2012) ‘Citizenship, Immigration, and the European Social Project: Rights and Obligations of Individuality’, The British Journal of Sociology 63(1): 1–21.

62.

Whelan

C.T.

Maître

(2007) ‘Measuring Material Deprivation with EU-SILC: Lessons from the Irish Survey’, European Societies 9(2): 147–73.

63.

World Bank (2011) Migration and Remittances Factbook 2011, 2nd edn. Washington, DC: The World Bank.