A progressive dilemma? Investigating cross-country variations in family-immigration policies through the lens of welfare-state regimes

ABSTRACT The notion of a ‘progressive dilemma’, according to which there is an intrinsic tension between comprehensive welfare states and large-scale immigration, has figured prominently in scholarly as well as political debates over the last decade. As one of the main categories of entry in most affluent democracies, family immigration stands out as a particularly interesting test case in this context. Building on this notion of a progressive dilemma, as well as on other theorizing on the welfare-migration nexus, this study examines whether the restrictive effects of certain risk factors on family-immigration policies, such as growing immigration and rising unemployment, have been conditioned by the type of welfare regime. The empirical analysis herein finds that increasing immigration and higher unemployment have triggered policy restrictions in Basic Security welfare states, but that the influence of these factors on policy changes is less clear in State Corporatist and Universal welfare states. Contrary to what the idea of the progressive dilemma would lead us to expect, Basic Security welfare states with weaker universal and redistributive features have been more likely to sharpen restrictions on the admission of family migrants when under pressure from increasing immigration and rising unemployment.


Introduction
International immigration has become a pivotal subject in contemporary academic and public debate about the welfare state.One important question here is whether or not there is a trade-off between large-scale immigration on the one hand and advanced, generous welfare states on the other.Much of the discussion surrounding this 'new progressive dilemma' 1 (Pearce 2004) has centred on the question of how modern welfare states can cope with increased immigration; however, it has also involved competing theories about the relationship between the design of welfare states and the openness of national immigration policies (Freeman 1986;Kymlicka 2015).On the one hand, some have argued that inclusive welfare states that offer a high level of social protection can be expected to be more reluctant to take in immigrants, due to fears of a weakened collective sense of shared belonging, as well as the greater public costs incurred (Andersen and Bjørklund 1990;Alesina and Glaeser 2004;Freeman 1995;Hay and Wincott 2012).Others, by contrast, have claimed that universal welfare states with extensive redistributive provisions tend instead, through their inclusive and egalitarian logic, to foster more open policies when it comes to immigrant admission and immigrants' rights (Boräng 2018;Sainsbury 2012).
Previous studies have challenged the idea of a progressive dilemma: e.g. by showing that immigration into European states has not affected welfare-state effort negatively (Fenwick 2019); that inclusive welfare states are not associated with higher levels of welfare chauvinism among the population (Van der Waal, De Koster, and Van Oorschot 2013; Mau and Burkhardt 2009); and that immigrants' rights have not been curtailed in countries with an inclusive welfare state (Römer 2017;Sainsbury 2012).Less attention, however, has been paid to the relationship between the welfare state and admission policies, i.e. immigration policies, with their regulation of entry and settlement.A few recent studies, however, have investigated the relationship between welfare states and cross-country variations in immigration policies.They show that more generous welfare states have been more favourable towards admission on humanitarian grounds (Boräng 2018), and that less generous welfare states have been more inclined to introduce skill-selective policies for labour immigration (Kolbe and Ayran 2019).
This paper advances research on the relationship between the welfare state and migration policy by utilizing a dynamic institutional model focused on the conditioning role of the welfare state.Previous studies have shown that tensions over immigration can harden in times of economic distress and of increasing immigration inflows, which in turn can affect admission regulations (Neumayer 2005;Money 1999).I argue that the degree to which these risk factors provoke restrictive changes in family-immigration policies depends on the institutional design of different welfare states.The idea is that, depending on their institutional design, different welfare states are more inclined or less to restrict family immigration when under pressure from increasing immigration and rising unemployment.Moreover, contrary to what the idea of the progressive dilemma would lead us to expect, leaner welfare states prove to be more inclined than generous ones to impose restrictive immigration polices when faced with growing pressures connected with large-scale immigration and rising unemployment.Conversely, welfare states with stronger universal and redistributive features tend to mitigate the associated tensions and thus to counteract restrictive policy reforms.
Furthermore, I argue in this paper, we must understand the relationship between the welfare state and variations in admission policies if we are to understand the development of family-immigration policies properly.As one of the main categories of entry in most affluent democracies, family migration has drawn increasing attention in recent years, both in political and in scholarly debate (Eggebø and Brekke 2019).Several studies have pointed to growing restrictions in family-immigration policies in numerous states during the last few decades (e.g.Bonjour and Kraler 2015).Others have stressed the existence of cross-country differences in policy configurations (Ahlén 2022).It is not just that countries differ in terms of restrictiveness; they have also introduced different types of policies with regard to eligibility criteria and qualifying conditions for resident sponsors and/or incoming family members.
Yet, comparative analyses of the determinants of this policy variation are largely lacking.Thus, we know very little about the causes of varying policy configurations across affluent democracies over time.This paper addresses this gap by considering the following questions: Are differences in family-immigration policies linked to differences between welfare states?What role does the institutional design of welfare states play relative to the presumed tendency of growing immigration and rising unemployment to call forth policy restrictions in this area?
The progressive dilemma and competing theories about the relationship between the welfare state and migration policy The emergence of new economic, demographic, and political realities in the globalized post-Cold War era have prompted debate over how advanced welfare states can cope with large-scale immigration and growing ethnic diversity (Freeman 1995;Koopmans 2010).Since the late 1990s, a key objective of immigration policies across advanced industrial democracies with ageing populations has been to attract immigrant workers with certain skills and presumed economic value, in order to tackling demographic change (Ruhs 2013).At the same time, both political parties and the public at large have increasingly sought to reduce the influx of 'costly' and unwanted migrants, including low-skilled workers, family migrants, and persons seeking protection (de Haas, Natter, and Vezzoli 2018).
Moreover, different hypotheses have been put forward about the relationship between the inclusiveness and generosity of welfare states on the one hand, and the openness of immigration and immigrant-integration regimes on the other.The concept of a progressive dilemma in this area is based on the notion of an intrinsic tension between growing immigration and the modern welfare state (Pearce 2004).Sometimes this tension is seen in ideational terms.The argument here is that the ethnic diversity resulting from growing immigration reduces collective feelings of solidarity and shared belonging, making it more difficult to maintain the inclusive and generous welfare state that was built up during times of cultural homogeneity (Andersen and Bjørklund 1990;Alesina and Glaeser 2004).
Interest-based versions of the concept point instead to resource constraints and to pressures on welfare programmes stemming from large-scale immigration, which are thought likely to generate greater tensions in states with far-reaching welfare programmes, especially when the latter are organized on a universal basis (Hay and Wincott 2012;Razin, Sadka, and Suwankiri 2011;Freeman 1995).Moreover, these tensions associated with immigration is assumed to worsen in times marked by economic pressures, such as high levels of unemployment.These are expected to increase the reluctance of welfare states to receive immigrants, due to the greater costs that so doing involves (Neumayer 2005).The assumption either way -from either the ideational or the interest-based perspective -is that welfare states with strong universal and redistributive features will be more inclined to impose stricter admission policies when faced with largescale immigration and/or economic distress, due to distribution conflicts, welfare chauvinism, and a fear of ethnic particularization.
These arguments have been challenged by scholars who claim that, on the contrary, inclusive and generous welfare states are associated with more open immigration regimes (Boräng 2018;Kymlicka 2015;Sainsbury 2012).According to these accounts, welfare states with strong universal and redistributive features foster large-scale solidarity and greater tolerance among the majority population towards outgroups, because they reduce income inequalities and moderate conflicts over redistribution.As will be discussed further below, the inclusive and egalitarian logic that animates more comprehensive welfare states is expected to lead to more permissive policies, in terms of both admission regulations and immigrants' rights.
Given the moral underpinnings and policy features of family immigration, it provides a particularly relevant area for assessing these competing views and for analysing the impact of the welfare state on cross-national policy variations.Family migration differs from other immigration tracks in many respects, especially since it involves immigrants who join persons who already reside in another country.That is, family immigration involves the 'claim[s] of insiders' (Block 2012, 37).Scholars have accordingly stressed that strict admission requirements -particularly qualifying conditions focused on the deservingness of resident sponsors -collide with the egalitarian norms inherent in comprehensive welfare states, such as universality and equal treatment (e.g.Borevi 2015).At the same time, the insider/outsider overlap entailed in family migration encourages the adoption of additional policy practices in order to control admission.Most importantly, requirements can be placed both on incoming family members and on resident sponsors (the latter, for example, may be required not to be dependent on social welfare).One key aspect of this 'double conditionality' is the fact that demanding integration conditionssuch as income and housing requirements aimed at resident sponsors -can be used to pursue goals of immigration control (Bech, Borevi, and Mouritsen 2017, 3).
If we reconnect the progressive dilemma with the different ideational and interestbased perspectives outlined in previous research on the relationship between the welfare state and immigration policy, a couple of competing propositions emerge.On the one hand, welfare states with strong universal and redistributive features may have stronger incentives to impose stricter policies to regulate family-related admission, on account of the threat that such immigration might be thought to pose to the sustainability of the system.On the other hand, comprehensive welfare states may be more resistant to restrictive trends; and leaner welfare states, with their weaker universal and redistributive profile, may be more likely to encourage the view that family immigration has a negative impact on welfare expenditure and social cohesion (cf.Korpi and Palme 2003).

How does the welfare state influence family-immigration policy?
Building on previous theorizing about the welfare-migration nexus, this paper develops an institutionally based dynamic framework for analysing the influence of welfare-state regimes on family-immigration policies.While recognizing that a variety of factors lie behind reforms in the area of migration policy, I argue that the institutional design of different welfare states shapes family-immigration policy, by conditioning the influence of exogenous risk factors that can trigger restrictive policy reforms.This approach has similarities with that taken in prior studies that have pointed to the importance of the welfare state in shaping political patterns and outcomes in connection with migration and integration (Boräng 2018;Crepaz 2008;Kolbe and Ayran 2019;Römer 2017;Sainsbury 2012).In this study, however, I advance this line of research further.Not only do I theorize how variations in migration policies are linked to differences between welfare states; I also explore how the institutional design of different welfare states influences the degree to which increased immigration and rising unemployment provoke restrictive changes in immigration policy.
My focus on the conditioning role of welfare-state regimes reflects institutional theory, which often portrays the relationship between different factors and policy outcomes as varying according to the institutional context (e.g.Pierson 2004;Thelen 1999).Moreover, if we aim not only to describe but also to explain variations across institutional settings, we must -as Thelen stresses (1999, 391) -specify the mechanisms that reinforce a particular path or trajectory.Aware of the difficult task of identifying the causal links between institutional regimes and policy output, the discussion that follows provide a first attempt to disentangle how different welfare-state regimes may condition the influence of exogenous risk factors on policies relating to the admission of family immigrants.
Rather than confirming the existence of the ideational and interest-based mechanisms underpinning the idea of a progressive dilemma, I provide theoretical arguments in support of a contrary claim -namely that market-oriented welfare states with weaker universal and redistributive features are more likely to sharpen restrictions on the admission of family migrants when under pressure from increasing immigration and rising unemployment.

Institutional differences between welfare-state regimes
Comparative research on the effects of policies and institutions follows two different paths.The first, the 'regime approach', is perhaps best exemplified by Esping-Andersen's (1990) typology of three types of welfare-state regime: the Liberal, the Conservative Corporatist, and the Social Democratic.The second is the 'variable approach' (Palme 2006), which eschews the use of country categories in favour of a set of variables that aim to capture different aspects or dimensions of policy.
Since my aim is to investigate the role of welfare states in conditioning the impact of assumed drivers of policy restrictiveness in the area of family migration, the regime approach is to be preferred.The regime approach involves classifying countries in a way that captures the kind of institutional variation that is relevant for the purpose of this study.It bears noting that previous studies vary in how they do this (cf.Esping-Andersen 1990; Ferragina and Seeleib-Kaiser 2011;Korpi and Palme 1998).By employing a regime approach, this study goes beyond the measurement of 'welfare state generosity' (Scruggs, Jahn, and Kuitto 2014) commonly used in studies of the relationship between the welfare state and migration policy (see, e.g.Boräng 2018;Römer 2017).Instead, I focus on the institutional configurations of different welfare states, in order to identify mechanisms that link institutional attributes of welfare states with different policies for family-based admission.
Focusing on institutional variations, I distinguish between regimes according to how they organize social protection and family policy: the Basic Security regime, the State Corporatist regime, and the Universal regime (cf.Korpi and Palme 1998) In the Basic Security welfare state, universal contributory flat-rate benefits predominate, or (in the case of Australia) income-tested ones.There is a strong emphasis on means-tested social assistance, not least in order to compensate for low social-insurance benefits.Earnings-related benefits are the main element in the State Corporatist regime, the purpose being to protect workers from loss of earnings.Accordingly, a very large part of the financing comes from social-security contributions.The Universal welfare state, finally, combines flat-rate benefits for those outside the labour market with earnings-related social-insurance provisions within a universal framework (Österman, Palme, and Ruhs 2019).
It is crucial to keep in mind that there will always be instances where a country deviates from the typical characteristics of its regime type.For example, Basic Security welfare states may have certain important redistributive elements as they focus resources on the poorest and may also provide access to universal services (Sainsbury 2012).Nevertheless, each regime encompasses a set of institutions that usually coexist and adhere to a shared logic, forming a distinctive 'package' of institutions (Österman, Palme, and Ruhs 2019).
When it comes to family policy, Korpi (2000) argues, three broad models can be found in Western welfare states.Convenient enough, these family policy models coincide with the welfare state regimes identified on the basis of the social protection systems in general.The market-oriented model, which provides a low level of public family support, has been introduced in Basic Security welfare states.The traditional breadwinner model, which provides social benefits in support of a gendered division of labour within families, characterizes the State Corporatist welfare regime.Universal welfare states, finally, have introduced a dual-earner model of family policy, with earnings-related parental leave and publicly subsidized childcare services (Bäckman and Ferrarini 2009).
Again, it is important to note that some countries deviate to some extent from the typical characteristics of these broader models.Specifically, previous research has shown that countries like Germany, the Netherlands, and the UK have taken steps towards a dual earner model by introducing paid leave schemes and increasing investments in childcare (Morgan 2013).Yet, despite the overall trend of expanding family policy in recent decades, significant cross-national variations exist among advanced welfare democracies, aligning with the distinctions between these broader models of family policy (see e.g.Bäckman and Ferrarini 2009).Furthermore, these variations hold particular significance for the period of interest in this study, spanning from 1985 to 2010.
On this basis I classify Australia, Canada, the United Kingdom, and the United States in the Basic Security regime/Market-oriented family model; Austria, Belgium, France, Germany, Italy, Luxemburg, and the Netherlands in the State Corporatist regime/Traditional breadwinner family model; and Denmark, Finland, Norway, and Sweden in the Universal regime/Dual earner family model.Ferragina and Seeleib-Kaiser's (2011) group my twelve European and three non-European countries into three different regime types in the same way but use different labels: the Anglo-Saxon liberal regime, the Continental European conservative regime, and the Nordic social democratic regime.This illustrates that the clustering is robust.Since I am interested in institutional variation, I use institutional labels instead of geographical ones: i.e. the Basic Security regime, the State Corporatist regime, and the Universal regime (see Österman, Palme, and Ruhs 2019).This is in line with a prominent critique of Esping-Andersen's regime typologythat it conflates the institutional aspects of social-protection programmes with the intended and unintended outcomes of the different institutional models (Korpi and Palme 1998).
In view of these institutional differences between welfare regimes, I expect three main mechanisms -deservingness, solidarity, and dependency -to shape the degree to which certain risk factors call forth restrictive changes in family-immigration policy.In the remainder of this section, I discuss differences between welfare-state regimes with regard to these mechanisms, building on both micro-and macro-level theories relating to the welfare-migration nexus.I propose four hypotheses pertaining to the impact of welfare-state design in this area.

Deservingness and solidarity with outsiders
Many scholars have argued that the programmatic design of social-welfare institutionsin terms of the degree of selectivity in benefit entitlement (is a given programme principally universal, work-based, or means-tested?), as well as the way in which benefits are constructed (to what extent are they designed to reduce income inequalities?)-shape the degree to which immigrants are perceived as 'deviant', and thus as more deserving of social support or less (Larsen 2008;Kumlin and Rothstein 2010;Reeskens and Van Oorschot 2012;Mårtensson et al. 2021;Sainsbury 2012).In this view, two key factors conditioning public perceptions of the deservingness of immigrants are, on the one hand, the degree of inclusiveness of welfare states; and, on the other, whether social-welfare institutions have been constructed in such a way as to exacerbate or to lessen conflicts over redistribution.
Earlier analyses of public attitudes on the social rights of immigrants have shown welfare chauvinism -the idea that immigrants' access to the welfare state should be limited and conditioned upon certain deservingness criteria -to be more prevalent in Basic Security welfare states, which are organized on a more selective basis and which provide flat-rate benefits at comparatively low levels (Mau and Burkhardt 2009;Van der Waal, De Koster, and Van Oorschot 2013).In these settings, immigrants are more likely to be regarded as competitors for scarce common resources, and thus as less deserving of redistributive benefits.In Universal welfare states, where entitlement is universal and benefits are so constructed as to result in a more equal distribution (Korpi and Palme 1998), citizens are less concerned about the potential negative impact of immigration upon welfare, and so are more favourable to the social inclusion of immigrants (Van der Waal, De Koster, and Van Oorschot 2013).Finally, citizens in State Corporatist welfare states -where work performance is deeply inscribed in the construction of social-welfare schemes -have been found to be more chauvinistic than their counterparts in Universal welfare states, but slightly less so than citizens in Basic Security ones (Van der Waal, De Koster, and Van Oorschot 2013).
A similar line of argument has underpinned a few cross-country studies on the macrolevel relationship between the welfare state and migration policy (Boräng 2018;Kolbe and Ayran 2019;Römer 2017;Sainsbury 2012).The idea that welfare regimes have spill-over effects on other policy areas has been called into question (cf.Kasza 2002); however, the studies mentioned above build on the view -prevalent within the welfare-migration literature -that institutional differences between welfare states shape the politics and policies of migration (Crepaz 2008;Geddes 2003).
Arguing that comprehensive welfare states foster large-scale solidarity and trust in a society, Boräng (2018) finds that such welfare states also apply more permissive admission policies on humanitarian immigration.Other observers emphasize a similar logic, according to which universal welfare states create a stronger sense of solidarity and a greater measure of tolerance vis-à-vis outgroups (Kymlicka 2015;Sainsbury 2012).Thus, in addition to explanations emphasizing cost-benefit concerns among citizens about the utility and deservingness of immigrants, there are arguments that stress the impact of feelings of solidarity (or their lack) on immigration policies -with such feelings being encouraged or discouraged according to the institutional design of the welfare state in question.
Linking these mechanisms with regime attributes like entitlement bases and benefit constructions, we can expect growing pressures in connection with large-scale immigration and higher unemployment to be more likely to trigger restrictive policy reforms in Basic Security welfare states, somewhat less likely in State Corporatist ones, and least likely in Universal welfare states.Such an outcome is the opposite of that predicted by the progressive dilemma, according to which exogenous risk factors will tend to elicit impose stricter migration policies in more comprehensive welfare states (Andersen and Bjørklund 1990;Alesina and Glaeser 2004;Freeman 1995;Hay and Wincott 2012;Razin, Sadka, and Suwankiri 2011).

Family-policy models and dependency
When it comes to the third mechanism, defined here as dependency, social family models are of particular relevance.Insights from the comparative welfare-state literature have shown the influence of the institutional design of social family policy on gender and family structures, as well as on gender-related outcomes -in connection, for instance, with income inequality and the division of labour (Bäckman and Ferrarini 2009;Korpi 2000;Palme 2006).
While expectations related specifically to social family policy or family migration are missing from previous research that asserts the existence of a progressive dilemma, a few earlier studies have discussed the interplay between the design of family-policy models and the politics and policies of family immigration (Borevi 2015;Eggebø 2010;Shutes 2016).Since the majority of immigrating spouses are women, one central line of argument in these studies is that the gendered perceptions and outcomes produced by different models reinforce different types of dependency associated with family immigration: i.e. dependency on work (the market), on one's partner/resident sponsor (family), or on social benefits (the state) (Eggebø 2010).
If we are to link these different family norms to family-immigration policies, we must first highlight the multidimensionality of admission requirements.Essentially, policies regulating the admission of family members can be divided into eligibility criteria (EC), which regulate who can apply for family immigration; and qualifying conditions (QC), which are additional requirements that resident sponsors and incoming family members must meet in order to obtain a given legal status.EC build on a circumstantial logic of restrictiveness; QC are behavioural qualifications that eligible applicants must fulfil in order to gain admission (Ahlén 2022).
Taking into account the differing dependencies inherent in different family-policy models, we could expect varying results from the different regime types when they are faced with pressure from presumed drivers of policy restrictiveness, like increased immigration and rising unemployment.In Basic Security welfare states, which have a large element of market-oriented family policies, family immigration is likely to be subject to more restrictive requirements overall, particularly in connection with educational qualifications and economic self-sufficiency -the idea being to promote participation in the labour market and to discourage dependence on welfare (Shutes 2016).State Corporatist welfare states, characterized by an emphasis on the traditional family and the role of male breadwinners, are also likely to impose stringent QC, especially for resident sponsors.In Universal welfare states, by contrast, we can expect QC -especially financial requirements aimed at resident sponsors -to be less stringent, inasmuch as such measures contradict the principle of autonomy inherent in the dual-earner model characteristic of these welfare states (Borevi 2015;Eggebø 2010).
The relationship between family-immigration policies and family-policy models can thus be expected to feed into the proposed relationship between family-immigration policies and welfare-state regimes.In addition, the dependency mechanisms entailed in different family-policy models can be expected to influence the type of admission policies applied in the area of family immigration, with countries that follow market-oriented and breadwinner-based family-policy models imposing more stringent QC for family immigration than countries with dual-earner models.

The impact of welfare-state regimes on the restrictive effects of risk factors
In view of the above discussion, we can expect differences between welfare-state regimes in terms of deservingness, solidarity, and dependency to influence the degree to which risk factors provoke restrictive changes in family-immigration policy.Drawing on findings of previous studies (e.g.Neumayer 2005; Money 1999), we can identify these risk factors as rising unemployment and growing immigration.
Previous studies have argued that states become more concerned about effectively protecting their borders and regulating immigration in times of resource scarcity and high unemployment (Helbling and Leblang 2019;Neumayer 2005;Gudbrandsen 2010).Since the unemployment rate is a strong indicator of tensions on the job market and of economic pressures overall, it is a plausible driver of immigration policy reform, and therefore a useful measure for analysing policy variations over time (Natter, Czaika, and de Haas 2020; Kolbe and Ayran 2019).Thus, it can be expected that states promote greater immigration policy restrictiveness to decrease pressure on fiscal resources when they perceive themselves as being overburdened by rising unemployment.Many have argued that inclusive and generous welfare states are likely to impose stricter immigration policies when faced with economic distress (Andersen and Bjørklund 1990;Alesina and Glaeser 2004;Freeman 1995;Hay and Wincott 2012).As argued above, however, other have shown that conflicts over redistribution are actually more visible in market-oriented welfare states, which are organized on a more selective basis and which provide less generous benefits (Sainsbury 2012).In such settings, immigrants are more likely to be regarded as competitors for scarce common resources, with 'undeserving' immigrants being pitted against 'deserving' nationals (Hemerijck et al. 2013).
Adherents of the idea of the progressive dilemma have stressed the concentration of immigrants in specific localities as a driving force behind restrictive admission policies (Freeman 1986;Money 1999).Others, however, have claimed that an influx of immigrants elicits a range of responses, depending on the structural and institutional setting (Boräng 2018;Van der Waal, De Koster, and Van Oorschot 2013).Mau and Burkhardt (2009) find that, in Western European states, attitudes in support of stricter policies are institutionally conditioned: they are not simple reactions to recent immigrant inflows or to the size of a foreign-born population in a country.The key factor, they accordingly argue, is whether social welfare is organized in such a way as to reinforce or to lessen the perception of 'us' vs 'them' when common welfare resources are at stake.
The type of welfare state may thus influence policy changes in the area of family immigration, by either mitigating or aggravating the impact of certain risk factors that tend to promote greater policy restrictiveness.In this view, the welfare state is an intervening institutional variable which conditions the relationship between presumed drivers of restrictiveness on the one hand and policy output on the other.However, in contrast to arguments that assert the existence of a progressive dilemma, the account presented above leads us to expect increasing immigration and rising unemployment to trigger policy restrictions most strongly in welfare states with weaker universal and redistributive features.Accordingly, I propose two hypotheses: H1: When unemployment is rising, restrictions in admission policy are more likely to be introduced in Basic Security welfare states than in State Corporatist or Universal ones, and more likely to be introduced in State Corporatist welfare states than in Universal ones.
H2: When immigration is increasing, restrictions in admission policy are more likely to be introduced in Basic Security welfare states than in State Corporatist or Universal ones, and more likely to be introduced in State Corporatist welfare states than in Universal ones.
In addition, following the reasoning in the previous sub section, the different family-policy models in the Basic Security and State Corporatist welfare states can be expected to pave the way for more restrictive QC.Universal welfare states, on the other hand, can be expected to show greater concern with the stratifying implications of behavioural requirements for admission, because such measures challenge the principles of inclusion and equality inherent in welfare states of this type.We have two complementary hypotheses here: H3: When unemployment is rising, stringent QC are more likely to be introduced in welfare states that follow market-oriented (Basic Security welfare states) and breadwinner-based (State Corporatist welfare states) family-policy models than in welfare states with dualearner models (Universal welfare states).
H4: When immigration is increasing, stringent QC are more likely to be introduced in welfare states that follow market-oriented and breadwinner-based family-policy models than in welfare states with dual-earner models.

Data and operationalization
My empirical analysis is based on a time-series cross-section (TSCS) dataset covering 15 welfare states from 1985 to 2010. 2 The main dependent variable, family-immigration policies, is measured as the degree of restrictiveness in a set of policy indicators gauging the admission of family immigrants.Restrictiveness is measured from 0-1, where higher values represent a higher degree of restrictiveness and lower values a lower one.
Measurements are taken from the Immigration Policies in Comparison (IMPIC) dataset, which provides cross-country data on migration policies for the 1980-2010 period (Bjerre et al. 2016).The aggregated variable for family-immigration policies includes the two dimensions of eligibility criteria (EC) and qualifying conditions (QC), each of which includes four policy indicators for resident sponsors or for incoming family members. 3I also carried out separate analyses for EC and for QC.The policy indicators yielding these measurements are described in Table A1 in the Appendix.
According to the theoretical expectations set out in the previous section, growing immigration and rising unemployment are risk factors that prompt policy restrictiveness in connection with the admission of family migrants.Immigration is measured as the yearly inflow of international immigrants per 1000 people (UNDESA 2015). 4International immigration as defined here includes any person who changes his or her country of usual residence for more than three months and who have been granted a valid residence permit in the country of destination. 5Unemployment rate is measured as the percentage of unemployed persons in the labour force (Dahlberg et al. 2020).
To control for party-political dynamics, the Ideological composition of governments index from the Party Government Dataset (Williams and Seki 2016) is included.Labelled Government orientation here, this index registers the relative strength of different parties -classified in left/right terms -in government and in parliament. 6The index runs from 1 to 5, where lower levels represent stronger right-party influence and higher levels represent stronger left-party influence.Natter, Czaika, and de Haas (2020) have shown that, across 21 Western states between 1970 and 2012, the effect of the political ideology of governments and parliaments on the overall restrictiveness of immigration reforms was limited.Others have argued that party politics, including left/right party divisions, have influenced reforms of immigration policy (e.g.Sainsbury 2012).Among these accounts, the dominant understanding is that centre-right parties as well as radical-right ones are more inclined to push for restrictive policy reforms than are centre-left parties (e.g.Bucken-Knapp, Hinnfors, and Spehar 2014; Gudbrandsen 2010).
The 2003 Directive on the right of family reunification (2003/86/EC) is included as a dummy variable, with member states that acceded to the Directive being coded 1 from 2003 on. 7Some previous accounts have emphasized the impact of the Directive on policy trends in EU member states.On the one hand, the Directive limits the leeway of member states to impose certain restrictions in EC, by establishing a set of minimum rights in connection with family reunification (Groenendijk and Strik 2018).On the other hand, it has been argued, the minimum standards stipulated by the Directive were more stringent than the policies already in place in many member states, pushing countries to introduce more restrictive measures (Bonjour and Vink 2013).
GDP per Capita is included to control for structural pressures and overall economic performance (Dahlberg et al. 2020).Moreover, in order to assess the impact of these explanatory variables on changes in family-immigration policies, I lag all independent variables 1 year, except for Government orientation and the Directive on the right to family reunification. 8

Descriptive findings
Before turning to the modelling and inferential analysis, let us take a look at Table A3 (in the Appendix) and Figure 1, which show how family-immigration policies have developed historically.Table A3 shows the average degree of restrictiveness of policies for the admission of family immigrants in the three regime types in 1985 and 2010.Figure 1 compares the policy configuration of 1985 with that of 2010.The scatterplots measure restrictiveness in eligibility criteria and in qualifying conditions, where eligibility criteria run along the X-axis and qualifying conditions along the Y-axis.
When comparing 1985 with 2010, we see that some movements took place.In particular, growing restrictiveness was evident in many countries, with the degree of it increasing on average from 0.19 in 1985 to 0.34 in 2010.In 1985, the four Basic Security welfare states (symbolized by blue circles) were positioned closer to the upper right-hand corner of the plot, thus displaying higher degrees of restrictiveness in both EC and QC. 9 Four of the seven State Corporatist welfare states (green circles) combined more restrictive QC with less restrictive EC (including Germany and the Netherlands, but not France). 10Among the Universal welfare states (red circles), finally, Denmark and Finland had low degrees of restrictiveness on both dimensions, while Norway scored high on QC and Sweden on EC. 11  While the restrictive pattern of Canada and the US was fairly similar in 2010 to that in 1985, both Australia and the UK increased their restrictiveness in both EC and QC. 12 The differences among the State Corporatist countries also increased.Belgium and Italy, although introducing some restrictions, maintained low degrees of restrictiveness on both dimensions.Austria, France, Germany, and the Netherlands increased their policy restrictiveness quite substantially (Luxembourg remained in the same position as before). 13All four of the Universal countries increased their degree of restrictiveness on QC (with Norway moving further than the others), while Finland and especially Denmark increased their restrictiveness on EC as well.Notes: Restrictiveness is measured from 0-1, where higher values symbolize a higher degree of restrictiveness and lower values a lower one.(Instead of being from 1985, the measurement for Austria is from 1992; and that for Australia is from 1989.This is due to missing data.) The regime variance in 1985 was more in line with the theoretical reasoning set out in the previous section, according to which the Basic Security welfare states would be the most likely to institute stricter admission policies on family immigration, followed by the State Corporatist and then by the Universal welfare states.An association was also expected between family-policy models and the level of stringency in QC, with marketoriented and breadwinner-based welfare states being more likely to impose stricter conditions regarding self-sufficiency and other behavioural patterns.However, policy changes during the 1990s and 2000s blurred these policy configurations to some extent.The Basic Security welfare states continued to apply more restrictive admission policies on average, but there were shifts across countries in the other two regime types during the period under study (see also Table A3 in the Appendix).The positions of the different countries are not set in stone, and parallel factors appear to have influenced policy reforms in different countries.
Thus, evidence of restrictive changes in family-immigration policies brings the question of an emerging progressive dilemma in advanced welfare states to the fore.Have we reached a tipping point, where inclusive and generous welfare states have become more inclined than other welfare states to impose restrictions when faced with increasing immigration and worsening economic distress?Alternatively, are such risk factors instead more likely, as hypothesized in the theoretical section, to prompt policy restrictiveness in welfare states with weaker universal and redistributive features?The findings in my crosssectional comparison call for more rigorous analysis of the drivers of cross-country variation in family-immigration policies.It is to that task I now turn.

Estimation strategy
To investigate changes in policy restrictiveness regarding family immigration during the 1985-2010 period, I use a time-series regression model with panel-corrected standard errors (XTPCSE).For analysing time-series cross-sectional data (TSCS), the combination of OLS regression estimates with panel-corrected standard errors (PCSEs) allows for accurate estimation of variability in the presence of temporally and spatially correlated errors, as well as heteroscedasticity (Beck and Katz 1995).All of the models here include countryfixed effects (FE), in order to control for stable country characteristics and to avoid omitted-variable bias (Allison 2009).Following Kropko and Kubinec's approach (2020), these models utilize one-way fixed effects of within-country variation, rather than twoway fixed effects (which also add year-fixed effects).According to Kropko and Kubinec, the two-way FE model produces estimates that are difficult to interpret, because it does not distinguish clearly between the two dimensions of case and time.Since my interest in this study is in how certain predictor variables have influenced within-country policy change during the period studied, rather than in how countries have changed relative to other countries at a particular point in time, I have used a country FE model rather than a two-way FE model.
In Table 1, Model 1 includes only immigration and unemployment.It tests the average effects of these variables on admission policies.Model 2 represents the main model specification.It includes strategic interactions between welfare-state regime and the predictor variables of immigration and unemployment.Reflecting the conditional hypotheses outlined in H1 and H2, the argument underpinning this model is that the impact on policy changes of increasing immigration and rising unemployment can be expected to vary depending on regime.The same specification was used for assessing the aggregated measurement of admission policies, as well as of QC (Model 3) and of EC (Model 4) separately.Since family-immigration policies are a slow-moving variable, the model also includes a lagged dependent variable, in order to control for past values of policy restrictiveness and to absorb the risk of serially correlated errors (Keele and Kelly 2006;Beck and Katz 2011).The resulting model can be specified as follows: Here, Y it is the degree of restrictiveness in family-immigration policies in country i at year t.β 1 U it-1 is the unemployment rate and β 2 I it-1 is immigration inflow.β 3 P it is government orientation; β 4 E it is the Directive on the right to family reunification; and β 5 G it-1 is GDP per Capita.β 8 -β 11 denotes the interaction terms between unemployment rate (U) and immigration inflow (I) on the one hand, and the State Corporatist regime type (C) and the Universal type (S) on the other.The Basic Security regime constitutes the reference group for the interaction variables.θY it-1 represents the lagged dependent variable, and σ i is the inclusion of country-fixed effects.∈ it is the error term.
To test the robustness of the main models and to control for the potential confounding effects of the other independent variables, models 5-7 in Table A4 (in the Appendix) interact welfare-state regime with GDP per Capita and Government orientation.These models thus add the interaction terms β 12 G it-1 ×C i + β 13 G it-1 ×S i + β 14 P it × C i + β 15 P it × S i to Equation (1).Model 8 mirrors Model 3 in Table 1; but instead of a 1-year lag, immigration inflow and unemployment rate are lagged 2 years (t -2), in order to account for delayed effects of these variables on policy changes.

Time-series cross-section analysis of changes in family-immigration policies
With various dependent variables and alternative model specifications, Table 1 presents the estimations of the regression analysis of admission policies in 15 welfare states during the 1985-2010 period.Model 1, which includes only immigration and unemployment, shows the immediate average effects of changes in these variables on changes in admission policies; while models 2-4 include control variables and interaction terms.
While the impact of rising unemployment on admission policies is small and statistically insignificant, that of growing immigration is larger, being significantly associated with admission policies across the 15 countries in Model 1.The immigration-coefficient in Model 1 is 0.00117, which in effect means that an increase in immigration rates by 1 person per 1000 population generates an increase in policy restrictiveness by 0.00117 points (scale 0-1).While this cannot be considered as a particularly large effect, it is not insubstantial.An increase of restrictiveness in admission policies by 0.0117 points roughly corresponds to a 0.1 point change in any of the policy indicators, which, for example, would be the outcome if a country went from having a post-entry language requirement for incoming family migrants to also require pre-entry language proficiency for obtaining admission on a family basis.This result supports the claim of some scholars regarding previous immigration inflows as a driving force behind restrictive admission policies (e.g.Freeman 1986;Money 1999).However, as discussed further below, the influx of immigrants has elicited different responses in different types of welfare state.
The effect sizes and statistical significance of the lagged dependent variables in models 1-4 show that the level of policy restrictiveness is heavily determined by its past level.Still, since the inclusion of the lagged dependent variable in combination with country-fixed effects isolates the actual change in a country over time, it strengthens the case that the independent variables have a significant influence on the outcome (Keele and Kelly 2006).A drawback, however, of including both country FE and lagged dependent variables is that they produce biased parameter estimates (Nickell 1981).However, this problem is particularly large in a context of small-̅ T, and it decreases with the number of observed periods.Since the sample in this study includes 334 observations between 1985 and 2010 ( ̅ T = 22.3), the incidence of a 'Nickell bias' may be considered small (Beck and Katz 2011).
Table 2 presents the estimated marginal effects of immigrant inflow and unemployment rate under the different regime types.Estimates are based on models 2-4.They show the marginal effects of immigrant inflow and unemployment rate on the three different measures of admission policies when the other independent variables are held at their average values.
The result for Model 2 confirms that, in the Basic Security regime, both increased immigration and higher unemployment are associated with growing restrictiveness in admission policies.The coefficients of the predictor variables are similar in size and statistically significant, at the 95 per cent level (immigration) and the 99 per cent level (unemployment).In the case of unemployment, for example, an increase of 1 per cent generates an increase in policy restrictiveness of 0.0029 points (on a scale of 0 to 1).In the State Corporatist and Universal regimes, by contrast, the predicted risk factors do not have any statistically significant positive impact on policy restrictiveness.In fact, higher immigration is associated with less restrictive admission policies in the case of the State Corporatist regime. 15 Thus, hypotheses 1 and 2 as such may be regarded as confirmed, insofar as the Basic Security welfare states have been more inclined to introduce restrictions in admission policy when faced with pressures from increased immigration and higher unemployment.These findings challenge the theories underpinning the idea of the progressive dilemma, according to which inclusive and generous welfare states are more likely to impose new restrictions on continued immigration when under pressure from large-scale immigration and serious economic distress (Andersen and Bjørklund 1990;Alesina and Glaeser 2004;Freeman 1995;Hay and Wincott 2012).These predictor factors seem instead to trigger restrictive policy reforms more strongly in market-oriented welfare states with weaker universal and redistributive features.This corroborates the theoretical reasoning outlined in this study: in leaner welfare states, with their less inclusive and generous institutional attributes, solidarity with outsiders tends to be weaker, and adverse judgements on the deservingness of foreigners are more likely to surface.In these settings, the tensions expected to accompany growing immigration and rising unemployment appear to reinforce the perception of 'us' vs 'them', with immigrants more likely to be viewed as deviant and undeserving when common welfare resources are at stake (Hemerijck et al. 2013).
Where the two sub-dimensions of admission policy are concerned, the results shows that the influence of immigration and of unemployment is particularly noticeable with regard to QC in the case of the Basic Security regime (see Model 3).This is in line with the findings of previous analyses, which have stressed the inclination of Basic Security welfare states to introduce various behavioural conditions in connection with family immigration, such as language tests and financial requirements (Kofman 2018;Shutes 2016).Analysing immigrants' rights in the UK and the US, Sainsbury (2012) claims that the growing 'conditionality of rights' in these countries reflects a contractual view, in which the responsibility of individuals to support both themselves and their family members has been increasingly held out as a justification for restricting immigrants' access both to welfare benefits and to various legal statuses (51).Accordingly, these market-oriented welfare states seem to be less concerned with the stratifying implications of demanding requirements in connection with family immigration (cf.Shutes 2016).The results of my analysis here indicate that the conditional and selective features of Basic Security welfare states overlap with growing conditionality in policies regulating the admission of family immigrants.
Hypotheses 3 and 4 as such may also be regarded as confirmed, insofar as the Basic Security welfare states, with their market-oriented family policies, have been more inclined to introduce restrictive QC when faced with increasing immigration and rising unemployment.However, while the trend in the Basic Security welfare states is in accordance with the expected pattern when QC and EC are analysed separately, that in the other two regimes is not.Both growing immigration and rising unemployment are associated with less restrictive QC in the State Corporatist regime (the association is statistically significant in the case of immigration).In the Universal regime, by contrast, the positive immigration-coefficient is quite large (0.0069), as well as statistically significant at the 90 per cent level.This suggests that growing immigration into the Universal countries may have been a driving force behind the introduction of policies of an increasingly demanding type.This result is thus more in line with the arguments supporting a progressive dilemma rather than with the theoretical reasoning underpinning hypothesis 3.
If we compare policy changes (and the drivers behind these) as between the Basic Security and the Universal welfare states, the result indicates that the impact of different welfare states might vary according to the type of risk factor in question.In the Basic Security welfare states, both external (immigration) and internal (unemployment) factors are linked with policy restrictions.In the Universal welfare states, however, restrictions in QC are associated with the external pressure of increased immigration, but not with the internal pressure of higher unemployment.In this view, comprehensive welfare states, with their stronger universal and redistributive profile, may be more resistant to interest-based concerns about immigration as an economic threat, but less resistant to ideational concerns about immigration as a threat to social cohesion.It may be fruitful in future research to look more closely at in how different interest-based and ideational perceptions of immigration are modified institutionally.
As shown in Figure 1, many of the State Corporatist countries introduced policy restrictions during the period studied.Contrary to my conditional hypotheses, however, these changes do not seem to have been triggered by growing immigration or by rising unemployment.An alternative theory, which has been discussed in previous studies (Mårtensson et al. 2021;Ruhs and Palme 2018;Reeskens and Van Oorschot 2012), is that the comparatively high degree of institutional reciprocity embodied in the work-based entitlements typical of State Corporatist welfare states may have the effect of mitigating adverse judgements on the deservingness of immigrants, and of counteracting negative attitudes towards immigration and the social incorporation of immigrants.This possible relationship -between the institutional reciprocity embodied in social-protection programmes on the one hand, and migration policies for various categories of immigrant on the other -merits further exploration.
Regarding the different impact of immigration inflow in the different regime types, it is also important to stress that the form of immigration (i.e. the relative size of different entry categories) varies between the countries included in the analysis.This variation is not accounted for here due to data limitations.One concern in this regard, for example, is that the EU member states have all had free movement from other EU member states.As a result, these countries may have adopted more restrictive measures to regulate the entry of non-EU family members, non-EU labour migrants, and refugees as a way to offset their limited control over the admission of EU citizens.In contrast, the non-EU/EEA countries in the analysis do not face the same policy constraints.It would be helpful to have policy data that overlap with data on immigrant inflows that distinguish between different categories of entry.This should make it possible to explore the determinants of policy reforms in greater detail.
Lastly, returning to Table 1, a few comments on the coefficients of the control variables are warranted.Model 2 shows that both higher GDP per capita and the Directive on the right to family reunification are associated with growing restrictiveness in admission policy.GDP per capita is also associated with restrictive policy changes in QC (Model 3), and the Directive with restrictions in EC (Model 4).On the one hand, the positive effect of GDP per capita reflects the overlapping of two different tendencies during the period under study: all countries in the sample experienced growth in GDP during the period, co-varying with increasingly stringent admission policies across countries.However, there are no significant regime-specific effects when GDP per capita is interacted with the regime variable (see Table A4).On the other hand, the positive effect of GDP per capita also indicates that economic growth by itself is not a safeguard against policy restrictiveness.
The positive effect on restrictiveness -EC in particular -of the EU Directive on the right to family reunification suggests that the Directive, by seeking to harmonize certain regulations for the admission of family migrants across member states, incentivized countries to introduce more stringent EC than they had previously had in place (Bonjour and Vink 2013).This concerns, e.g. the minimum-age requirement.According to the Directive, member states may set a minimum age for both partners of no more than 21 years.As I have discussed in another study (Ahlén 2022), many countries had previously allowed partners over the age of 18 to immigrate on a family basis, so the threshold set by the Directive seems to have incentivized countries to adopt higher age requirements.
The coefficient of Government orientation is negative throughout (except in Model 4), indicating that a higher share of left-parties in government and in parliament is associated with less restrictive policies.This finding is in line with the view that parties of the centre right and of the radical right are more inclined to push for restrictive policy reforms than are centre-left parties (e.g.Bucken-Knapp, Hinnfors, and Spehar 2014; cf.Korpi and Palme 1998).However, this relationship is not statistically significant.This supports the conclusion of previous analyses that the effect in this area of the political ideology of governments and parliaments has been limited in Western states (Natter, Czaika, and de Haas 2020).

Concluding remarks
This study contributes to research on the relationship between the welfare state and migration policy by advancing a dynamic institutional framework, and by presenting empirical evidence of changes in family-immigration policies in 15 welfare states between 1985 and 2010.Four conditional hypotheses are elaborated, based on a combination of micro-and macro-level theories from comparative migration research and comparative welfare-state research.The theoretical framework proposes that the institutional design of different welfare states conditions the effect of risk factors on policy restrictiveness, and likely shapes family-immigration policy thereby.
My empirical analysis has shown that increasing immigration and rising unemployment triggered policy restrictions on the admission of family migrants in the Basic Security welfare states.These findings challenge the assumptions underlying the theory of an emerging progressive dilemma in affluent welfare states, according to which growing pressures in connection with large-scale immigration and serious economic distress increase the incentive to impose stricter migration policies in comprehensive welfare states (Andersen and Bjørklund 1990;Alesina and Glaeser 2004;Freeman 1995;Hay and Wincott 2012;Razin, Sadka, and Suwankiri 2011).As this study has demonstrated, however, this pattern has instead made its appearance in market-oriented welfare states with weaker universal and redistributive features.
In the State Corporatist and Universal regimes, by contrast, the predicted risk factors are not associated with policy restrictiveness, with one important exception: the case of the relationship between immigration inflow and QC in the Universal regime.This would seem to indicate that other factors prompted the introduction of restrictive admission policies in many of these welfare states.For example, the Universal countries have all experienced growing polarization in public opinion and rising success on the part of rightwing populist parties -developments linked in considerable measure to the framing of immigration in public debate as a threat to social cohesion and to the welfare state (Bech, Borevi, and Mouritsen 2017;Eggebø and Brekke 2019).Accordingly, the framing of a progressive dilemma among political parties and other political actors -and the extent to which this might influence on the development of migration policiesshould be explored more thoroughly in future research.
Another venue for future research relates to regional and international frameworks.My results indicate that the Directive on the right to family reunification triggered EU member states to introduce restrictive EC for the admission of family migrants.Moreover, calls for stronger international coordination intensified following the 'migration crisis' of 2015; and new initiatives, such as the Pact on Migration and Asylum released recently by the European Commission, may prove to have major implications for national policy practices, including in connection with family immigration.
Recent policy reforms regarding family migration have been described in previous studies (see, e.g.Bonjour and Kraler 2015;Bech, Borevi, and Mouritsen 2017;Eggebø and Brekke 2019;Kofman 2018), but systematic analyses have so far been lacking of the causes of policy variation in this area.This study presents a first comprehensive cross-country analysis of the determinants of family-immigration policies over time.While the causal effect of the institutional regime on policies is difficult to identify in a comparative setting (see, e.g.Cronert 2018), I have provided evidence in this paper that the admission of family immigrants represents another area where the institutional design of different welfare states interlinks with migration management in modern welfare states.As such, it contributes to welfare-state research and to comparative migration research, by demonstrating that institutional variations in welfare states affect not just immigrant policies, but immigration policies as well (cf.Sainsbury 2012;Koopmans 2010;Römer 2017).This study offers a new understanding of how the welfare state may condition the relationship between presumed drivers of restrictiveness in migration policy on the one hand and policy output on the other, thereby advancing research on the relationship between the welfare state and migration policy (cf.Boräng 2018;Crepaz 2008;Geddes 2003;Freeman 1986).

Notes
1.The 'new liberal dilemma' is another term that has been used to describe the tension between immigration and the welfare state (see e.g., Reeskens and Van Oorschot 2012).2. I chose this time frame for reasons of data availability.The upper limit (2010) is set according to the IMPIC dataset, and the lower limit (1985) in order to avoid too many missing values, particularly for immigration rates (UNDESA 2015) and family-immigration policies (Bjerre et al. 2016).Descriptive statistics for all variables are presented in Table A2 in the Appendix.3. Given that my aim is to investigate immigration policy -i.e., policies affecting the admission of family immigrants -the measurement used in this study does not relate to internal policies aimed at family immigrants, such as security of status or rights associated therewith.4. I focus on yearly inflows rather than on immigrant stocks, in order to capture the more recent dynamics connected with immigration inflows into various welfare states.The stock of foreign-born persons is largely dependent on the migration history of a country further back in time.I would argue that, for analysing the drivers of policy changes during a specific period (as in this case), recent immigration flows are a more adequate measurement.Since available data on immigration inflows that covers the studied period  does not distinguish between forms of entry (e.g.asylum, work, and family etc.), it is not possible to conduct any cross-country analysis of differences between these entry categories (see UNDESA 2015). 5. Citizens from the EU, EEA, and Switzerland who move to another European country are counted as immigrants if they have been granted a right of residence.However, the required duration of stay varies between countries (see UNDESA 2015).Here it is important to note that differences in national definitions and methods of data collection pose challenges when it comes to comparability.Specifically, countries that collect and publish data on international migration flows employ varying criteria to identify immigrants.This includes variations in terms of the definition of residence and the required duration of stay (see UNDESA 2015).Thus, effects of immigration inflows should be interpreted with some caution.6.It should be noted that the group of right-wing parties in this index includes radical-right parties with explicit anti-immigration agendas, such as the Freedom Party of Austria (FPÖ), the Northern League in Italy, and Freedom Party (PVV) in the Netherlands (see Williams and Seki 2016).7. Denmark and the UK are exempted from the common rules set out in the Directive (Groenendijk and Strik 2018).8.I refrain from lagging Government orientation, in order to avoid the risk of incorrectly associating policy changes with a former party-political dynamic in government and in parliament if a national election had been held during the previous year.9. Average degree of restrictiveness in the Basic Security welfare states 1985: Admission policies: 0.33, QC: 0.34, and EC: 0.33.10.Average degree of restrictiveness in the State Corporatist welfare states 1985: Admission policies: 0.14, QC: 0.2, and EC: 0.08.11.Average degree of restrictiveness in the Universal welfare states 1985: Admission policies: 0.13, QC: 0.1, and EC: 0.17.12. Average degree of restrictiveness in the Basic Security welfare states 2010: Admission policies: 0.43, QC: 0.47, and EC: 0.39.13.Average degree of restrictiveness in the State Corporatist welfare states 2010: Admission policies: 0.29, QC: 0.38, and EC: 0.2.14.Average degree of restrictiveness in the Universal welfare states 2010: Admission policies: 0.33, QC: 0.37, and EC: 0.29.15.Adding interaction terms between regime and the control variables to the models does not affect this result (see models 5-7 in Table A4).Similar tendencies can be discerned when immigration and unemployment are lagged two years (see Model 8 in Table A4).Increased immigration and higher unemployment are associated with growing policy restrictiveness in the Basic Security welfare states.However, the relationship is statistically significant only in the case of unemployment.In contrast, none of the predicted risk factors have a significant impact on policy restrictiveness in the State Corporatist and Universal welfare states. 14

Figure 1 .
Figure 1.Configurational model of policies for the admission of family immigrants in 15 welfare states, 1985 and 2010.

Table 1 .
XTPCSE models predicting three different measurements of admission policies regarding family immigration in 15 welfare states, 1985-2010.

Table 2 .
Average marginal effect (AME) of immigrant inflow and unemployment rate (per 1000 population) on three measurements of family-immigration policies in three welfare-state regime types.Effects of one unit change in immigrant inflow and unemployment rate (per 1000 population), with all other vari- ables held at their mean values.Observations per model: 334.Welfare-state regime types: Basic Security (AU, CA, UK, US), State Corporatist (AT, BE, DE, FR, IT, LU, NL), Universal (DK, FI, NO, SE).*

Table A4 .
Continued.Time-series regression with country-fixed effects and panel-corrected standard errors.Family-immigration policies are measured from 0 to 1, where higher values symbolize a higher degree of restrictiveness and lower values a lower one.Welfare-regime clusters: Basic Security (AU, CA, UK, US), State Corporatist (AT, BE, DE, FR, IT, LU, NL), Universal (DK, FI, NO, SE).