Introduction

Whereas migration studies in the 1980s and 1990s focused primarily on then national level, it has now increasingly turned attention to the role of the local level. The early preoccupation with national migrant integration policies led to criticism on migration studies for its reification of ‘national models’ (Bertossi 2011: Favell 1998) and even for ‘methodological nationalism’ (Wimmer and Glick Schiller 2002). Yet, there is a growing body of literature on the local dimension of integration policies (Jones-Correa 2001; Glick Schiller and Çağlar 2009; Caponio and Borkert 2010). This has contributed to a better understanding of how and why local integration policies develop in response to a specific local problem, policy and political circumstances (Alexander 2007; Garbaye 2005; Mahnig 2004; Zapata Barrero 2015).

However, thus far, very few studies have focused on vertical relations between policy levels and the implications these have for integration policies. This speaks first of all to the literature on multi-level governance as a way or organizing relations between different levels (Bache and Flinders 2004; Hooghe and Marks 2001; Piattoni 2010). With the recognition of not only a local turn in integration policies but also a certain resilience of national policies and even a marginal but growing presence of EU and regional policies, migrant integration has clearly evolved into a complex multi-level governance field. Whereas studies of national models often assumed clear top-down mechanisms for coordinating vertical relations between the national and local levels, other studies have revealed much more ‘entrepreneurship’ of local governments not only in developing their own integration policies but also in setting their ideas on the agenda of national governments and sometimes even the EU (Scholten 2013). Furthermore, Bak Jørgensen has identified various forms of ‘horizontal’ cooperation between cities across Europe as an important factor in vertical mobilization of cities for setting policy ideas on national and European agendas. To capture these various governance configurations in multi-level settings, various ideal types can be distinguished, including centralism, localism, decoupling as well as ‘multi-level governance’ (Scholten 2013).

However, rather than accounting for the development of different configurations (see my earlier work Scholten 2011), this article focuses on the consequences of these configurations for integration policies. In the past, centralist modes of policy coordination have had a strong effect in terms of creating policy convergence around a specific national model of integration. Now, we know that the local level has become increasingly prominent, what are the consequences of national-local relations in terms of creating either policy convergence or divergence? Some scholars have drawn attention to a growing discrepancy between national and local policies (Bak Jørgensen 2012) due to the decoupling of vertical relations. This may bring about policy fragmentation and even contradictory policies that convey very different policy messages to specific target groups and, sometimes, even involve open policy conflict between levels. Others have shown how new forms of convergence emerge in response to ‘vertical venue shopping’ by local governments or the establishment of ‘multi-level governance’ structures. This way, local politics, policies and problem developments would factor into integration policies alongside national factors.

This article asks how and why different configurations of relations between national and local governments affect the governance of migrant integration. It hypothesizes that different ‘vertical’ governance configurations will have an effect in terms of producing either convergence or divergence in the framing of migrant integration in policies at different levels. Following an embedded dissimilar case study design, The Netherlands and the UK are selected as two case studies because of differences in the structure of relations between national and local integration policies in both countries. The Netherlands is one of the countries that has been associated traditionally with a state-centric approach to migrant integration but has clearly witnessed a ‘local turn’ in terms of increasing involvement of the local level (Scholten 2013). In contrast, policies no migrant integration in the UK have traditionally evolved much more from the local level (Hansen 2000; Favell 1998; Spencer 2011), with a much more marginal involvement of central government. In order to capture national-local relations, the analysis focuses on two cities per country that are known for their different integration approaches: London (boroughs Tower Hamlets and Enfield) and Glasgow, and Rotterdam and Amsterdam. The analysis is based on a qualitative analysis of vertical relations based on document analysis and interviews with national as well as local policy actors and a quantitative mapping of policy frames in the two countries and the four cities.

Between National Models and Multi-Level Governance of Migrant Integration

Our understanding of migrant integration policies has for a long time been framed by the idea of ‘national models of integration’ (Thränhardt and Bommes 2010; Favell 1998, Brubaker). This would involve historically developed and nationally embedded ways of defining and acting upon migrant integration (Bertossi 2011). For instance, the French approach to migrant integration would be defined by its typical national Republican model, the UK approach by its historically developed race relations approach and the Netherlands by its multiculturalist approach inspired by the peculiar Dutch history of pillarization. Inspired by an historical institutionalist perspective, the idea of national models draws attention to what Favell ((1998): 47) describes as ‘the exclusively internal national political dynamics’ of integration policymaking and the ‘self-sufficiency’ in integration debates in many European countries, at least until well into the 1990s and early 2000s.

The idea of national models has also had a strong impact on migration studies in general and the study of integration policies in particular (Bertossi 2011, Bowen 2007). Not only due to its preoccupation with this national dynamics of integration policymaking but also due to the strong policy orientation in its early years of development, migration studies would have developed a form of ‘methodological nationalism’ (Wimmer and Glick Schiller 2002). This would have hampered the development of a better theoretical understanding of policy dynamics due to ‘context dependency and insufficient clarification on the conditions of generalizability’ (Thränhardt and Bommes 2010: 10). Furthermore, various scholars have highlighted how the idea of national models, sometimes reified by scholars, have played a profound role in national policy dynamics as well (Bowen 2007, Bertossi 2011, Duyvendak and Scholten 2012).

Rather than taking national models as a conceptual starting point, this article empirically questions the idea of a national model from a multi-level perspective. Speaking to the evolving literature on the local turn in migration studies (Jones-Correa 2001; Glick Schiller and Çağlar 2009; Alexander 2007; Caponio and Borkert 2010; Zapata Barrero 2015), it asks to what extent, and why, there is convergence or divergence in how migrant integration is framed at the national and local levels. It hypothesizes that there is a relation between the coordination structure for vertical relations between governments and the extent to which national and local governments will actually converge or diverge in their framing of migrant integration. Therefore, I will first elaborate the conceptual framework used for studying vertical relations between national and local governments and then connect this to literature on how migrant integration policies are framed nationally and locally.

Models of Multi-Level Governance

With the growing involvement of the local level (and to some extent the European level and in some cases regional levels as well), migrant integration has clearly evolved into a multi-level topic. Elaborating on earlier work on multi-level governance (Bache and Flinders 2004; Hooghe and Marks 2001; Piattoni 2010), Scholten (2013) has elaborated a typology of governance configurations in multi-level settings. Whereas theories on multi-level governance have been applied primarily to EU-national relations, this typology is particularly fit for the study of national-local relations. Furthermore, this typology stresses that ‘multi-level governance’ is only one of several ideal-type configurations that are to be found in multi-level settings. It should be distinguished from other types like top-down (or centralist), localist or decoupled configurations of relations between government levels. Firstly, the centralist ideal type is based on the idea of a clear top-down and hierarchical relations between levels of government. In a multi-level setting, this involves a clear central codification of the division of labour between levels and control mechanisms to make sure that policy implementation at the local level clearly follows central rules and reflects the central policy frame. In this type, one should expect there to be a clear national structure for policy coordination, such as a specialized department and a political responsible minister specific for integration policy.

Secondly, the localist type involves a more bottom-up perspective on governance in multi-level settings. In this type, policy competencies follow a principle of subsidiarity; what can be done locally should be done locally. In this type, local governments are involved in more than policy implementation; agenda setting and policy formulation also take places at the local level, in response to specific local circumstances. Local governments do more than just policy implementation, they formulate policies locally, respond to local policy agendas and exchange knowledge and information horizontally with other local governments.

In contrast to the often abstract use of the concept of multi-level governance, in this typology, it refers only to those situations where there is vertical interaction and joint coordination of relations between various levels of government. In contrast to the localist and centralist types, this should not involve a sense of hierarchy; rather, actors from various levels are to meet on a level playing field. There have to be ‘vertical venues’ for governments from different levels jointly to engage in meaningful policy coordination, such as forums or networks where organizations from different levels meet.

The final type refers to a situation in which vertical relations are absent or when there is ‘decoupling’. This means that in a single policy domain, there may be policies at different levels that are dissociated and may in fact even been contradictory. Evidently, this type can lead not only to policy conflicts between government levels but also to conflicting policy messages to the policy target groups and diminish policy effectiveness.

Convergence or Divergence of Integration Policy Frames in Multi-Level Settings

This article connects the literature on multi-level governance to that on the framing of integration policies, in an effort to deepen our empirical and theoretical understanding of the convergence or divergence of integration policy frames in multi-level settings. I prefer to use the concept ‘frames’ rather than ‘models’ as this captures the more constructivist way of how integration policies are defined, rather than taking these policies as products of historical circumstances. The concept of ‘policy frames’ as developed amongst others by Schon and Rein (1994) captures how policies provide ‘a way of selecting, organizing, interpreting and making sense of a complex reality to provide guideposts for knowing, analysing, persuading and acting’ (ibid: 32).

This article focuses in particular on how relations between levels of governments lead to frame convergence or divergence. In this context, the notion of resonance of frames or ‘frame alignment’ comes in (Snow and Benford 1988). This means multiple actors, or as in this research, multiple levels gradually adopt a similar frame (or mutually adjust their frames) in response to interaction and learning. This interaction can be social or political of nature and can involve mutual ‘horizontal’ interactions as well as more vertically coordinated interactions. While assuming that multiple frames can be found on each level, frame alignment would then mean that amongst these various frames, one becomes more prominent in response to interactions with other actors. The other way around, if no frame alignment takes place, one may expect frame divergence or possibly even frame conflict.

Applying this notion of frame alignment to migrant integration, the idea of ‘national models of integration’ can be seen as a form of top-down alignment of integration policies around a specific frame of integration. Connecting this to the multi-level typology, we can expect that if there is a strong centralist coordination structure, national and local policies will converge or ‘align’ in the way predicated by the idea of national models. Developments in the national framing are then to be followed by a similar reframing of local policies. Also, local frames are expected to be similar regardless of the specifics of the local (political/problem/policy) setting.

However, speaking to the broader literature on local integration policies, various scholars have predicated rather that there would be a local model of integration. In particular, local governments would be more accommodative towards cultural diversity in order to allow for pragmatic modes of problem solving at the urban level (Caponio and Borkert 2010, De Zwart 2005, Vermeulen and Stotijn 2010). This means that local frames will be similar across various localities but different from the national level as local governments are often confronted with integration problems in different ways than on the national level. Thus, this localist type is most likely to be associated with divergence between national and local level policies and convergence between policies in different localities. In fact, as Bak Jørgensen (2012) shows, such ‘horizontal’ convergences between localities can be reinforced by various forms of interaction and network formation between cities.

This differs again from other scholars who have rejected the idea of local (as well as national) models of integration, arguing instead that characteristics of different localities tend to be so different that horizontal alignment in terms of a local model of integration would be highly unlikely. This can refer to not only differences in terms of the local political setting (Mahnig 2004; Garbaye 2005) but also different local economic circumstances (Glick Schiller and Çağlar 2009) or different migration histories. Rather than resulting in one dominant local frame, this is likely to a great variety of frames in different local settings. On a vertical dimension, this may be associated with decoupling in the absence of vertical relations, contributing to the fragmentation of integration policies in a multi-level setting. However, some degree of convergence on the vertical dimension could be produced by the interaction between levels of government. Typically, this is likely to produce a division of labour in which there is convergence on specific policy elements while allowing policies to adjust to local settings in specific ways.

The relations between governance configurations in multi-level settings and frame divergence or convergence as hypothesized in the literature and to be examined in this article are brought together in Fig. 1.

Fig. 1
figure 1

Expected association between types of governance configurations in multi-level settings and convergence and divergence in policy frames on the national and the local levels

Methods

This analysis follows an embedded dissimilar case study design. It focuses on two countries, the UK and the Netherlands, that are known to have different governance structures in the field of migrant integration policies. Although both unitary states, the Netherlands has traditionally followed a state-centric approach, whereas UK policies in the field of migrant integration have traditionally focused much more on the local (or even borough) level. Within both countries, two cities were selected that are known to have different approaches towards migrant integration and that do not have a strong mutual relation in order to avoid contamination with the variable of policy framing. In the Netherlands, Amsterdam and Rotterdam were selected as the two largest cities where migrant integration was an important policy concern. In the UK, the selection is complicated by the importance of devolved authorities. Hence, one Scottish city was selected, Glasgow, as well as one English city, London. For the former, an analysis was made of the Scottish level as well. For the latter, to do justice to the complex governance situation within London, the analysis focused on two London boroughs (that differed in terms of type of migrant population and political orientation of the borough council), Tower Hamlets and Enfield, and on the Greater London Authority. Hence, in total, the research spanned two national policy arenas (the Netherlands and the UK), two Dutch cities (Amsterdam and Rotterdam) and five subnational authorities in the UK (Scotland, Glasgow, GLA, Tower Hamlets and Enfield).

In terms of methods, this research relies primarily on a qualitative analysis of policy framing and multi-level governance relations and a quantitative mapping of different policy frames. First, the main policy frames on the national as well as the local level were identified inductively by analysing national and local policy documents and a review of available secondary literature. This inductive approach also means that in both countries, different frames were identified. However, the key interest of this article is on the convergence or divergence of frames between the national and the local level rather than on a cross-case comparison of the content of these frames.

Secondly, an analysis was made of the quantity of attention in national and local policy agendas to these frames identified in the first step of the qualitative analysis. This was done by the coding of government documents on the agenda of national parliament or local city or borough councils. This coding was based on an operationalization of the frames identified in the first step into Boolean word searches, which were entered into digitalized archives of parliamentary or council records. Footnote 1 A strength of this mapping is that using similar key word strings (Boolean searches) provides a good comparative method for studying when and how much a specific frame is mobilized in specific national and local policy contexts.Footnote 2 A weakness is that due to different archival methods (and plain differences in the overall number of policy documents discussed in different councils or parliaments), a comparison in absolute numbers between different government entities is not possible. However, this method enables an analysis of the relative prominence of the different frames as well as an analysis of trends in the amount of attention attributed to a specific frame. Thus, the cases can be meaningfully compared in terms of these trends and relative prominence of frames, which is the main focus of analysis. Most cases allowed for an analysis of digital archives starting from 2001, allowing the analysis to cover a period of 10 years (2001–2011).

Third, a qualitative in-depth analysis was made of governance relations between the various levels of government. This involved interviews with key policymakers and experts who have been or are still directly involved in national and/or local policymaking in this domain. In the UK, 12 interviews were held (with experts, with national policymakers and local policymakers) and in the Netherlands, 15 (with local policymakers, with national policymakers and with experts). These semi-structured interviews were essential for tracking precisely the mechanisms and relations that characterize the four types of governance relations in multi-level settings that have been described above.

Migrant Integration Policies in the UK

Although policies in the domain of migrant integration have been framed very differently than in various other European countries, for instance, averting the use of the concept integration until recently, the UK has established policies in this domain. Following the UK complex multi-level governance system, an analysis will be made of national policies as well as well as Scottish policies, policies in the two London boroughs Tower Hamlets and Enfield and in Glasgow.

National UK Level

Policies on migrant integration in the UK are rarely framed explicitly in terms of integration. Furthermore, they also lacked a clear central coordination structure or even a clearly formulated central policy philosophy. Favell (1998: 96) speaks in this respect of ‘calculated, piecemeal, evolutionary, anti-philosophical pragmatism’. UK policies emerged in a mostly ad hoc manner in response to local demands and experience (Rose 1969). A review of secondary literature reveals that the UK policies until well into the 1990s were inspired by two distinct policy frames (Spencer 2011: Bleich 2002). First, a race relations frame that focuses primarily on relations between ethnic and racial communities, for instance, on promoting inter-ethnic contact, preventing residential segregation and building inter-community trust. Second, a race equality frame that focuses much more on the legal dimension of integration, such as equal rights and anti-discrimination. Studies show that this race equality frame has been dominant in national policy developments (Cantle 2005), whereas on the local level, there was much more attention to race relations (interview with UK expert). This reflects a clear division of labour between national and local governments, with the former focusing on discrimination and equality and the latter more on cohesion and prevention of tensions and unrest.

This situation of pragmatic localism changed in the early 2000s in response to a series of riots in northwestern cities, the so-called Milltown riots (Bradford, Burnley and Oldham), that put migrant integration framed now as ‘community cohesion’ on the national agenda (McGhee 2008). In response, several government committees were put to work, of which most notably the Cantle committee that issued its report in 2002. This report recommended the development of a community cohesion policy primarily oriented at promoting relations between minority groups, building bridging social capital, preventing parallel lives and developing a shared sense of ‘Britishness’ (Cantle 2001). In response to the Cantle Report, the Home Office set up a Community Cohesion Unit that was to coordinate new policy measures. Later, this committee was repositioned to the Department of Communities and Local Government, which was in a better position to coordinate the new policy plans with the local level. However, the interviews indicate that the new national coordination structure remained modest. Central government had little means for top-down policy coordination. It did however develop several soft government mechanisms that proved rather effective, such as the Pathfinder and Beacon Status programs that awarded specific local cases as best practices to set an example for other local governments (interview with national and local policymakers). Furthermore, the coordination on the national level between various departments lacked ‘horizontal integration.’ As Spencer (2011) shows, there were discrepancies between the policy arguments of the Home Office and the Department of Communities and Local Government, which Spencer describes as persistent ‘policy silos’ that complicate the development of a coherent integration policy.

Importantly, this national agenda setting of ‘community cohesion’ and establishment of a modestly more centralist approach (interview with UK policymaker), with a central role for the Home Office. This was evident amongst others in its 2005 memorandum ‘Improving Opportunity, Strengthening Society’, containing a clear national perspective on race equality and community cohesion. This had important implications in terms of the ‘framing’ of migrant integration. On the national level, ‘community cohesion’ became the dominant frame (see Fig. 2a) and gradually also on the local level (2B, 2C and 2D). Notably, this frame did not emerge as prominent in Scotland and Glasgow, at least until very recently.

Fig. 2
figure 2

Policy frames in UK national, Scottish and local (Glasgow, GLA, Enfield, Tower Hamlets) migrant integration policies (annual number of documents on the agenda in the period 2001–2011)

Interviews confirm that the community cohesion frame builds in several respects on the race relations frame, although shifting attention from specific racial minorities to cohesion of society at large. The community cohesion frame stated that promoting community cohesion would be the solution to the problem of growing segregation, while blaming past policies focusing too much on just race equality and underestimating the importance of intergroup relations. Consequently, one can only speak of a partial alignment between local and national race relations frames in response to the Milltown riots; the race relations became more prominent at both levels in this period but were also reframed significantly into the community cohesion frame. Furthermore, this was not a case of top-down alignment, but rather a product of exchanges of ideas and experiences between both levels via in particular the Cantle committee.

Another key focus event in the development of British policies was the 2005 bombings in London. In the aftermath of the London bombings, the Department of Communities and Local Government established a Commission on Integration and Cohesion (CIC 2007) to provide advice on the national and local integration and cohesion policies, bringing together experts from the national as well as the local level. The CIC differed from the post-Cantle policy initiatives in much less trying to develop a single master plan for community cohesion policies. Rather, it reconfirmed that community cohesion policies should be primarily local policies, and there should be no ‘one size fits all’ (ibid: 57). Indeed, the quantitative mapping of policy frames clearly shows that the community cohesion frame became the dominant frame not only in national policymaking after 2005 (see Fig. 2a) but also at the local level (see 2C and 2D, besides Scotland and Glasgow, and only very briefly in GLA). This confirms that frame alignment did take place between the national and local levels around the community cohesion frame in response to mutual interactions very much reflecting the multi-level governance type, with another ‘vertical’ commission, the CIC, as the main venue for multi-level interaction.

Furthermore, the CIC was the first formal government committee to coin the term ‘integration’ into British discourse. However, it refrained from providing a single definition of integration, leaving space for different definitions that match local conditions (Commission on Integration and Cohesion 2007: 39). Figure 2a does show that, following CIC, ‘integration’ was more used in national policy discourse than before but only for a brief period and never really to the extent as community cohesion played a role in national policy discourse. The 2012 memorandum ‘Creating the conditions for integration’, published by the department of CLG under the government coalition led by David Cameron, provided a new indication that the coordination framework at the national level has remained weak. In this respect, though an integration frame did emerge in national policies, it was never effectively substantiated on the level of national policy measures nor a national policy coordination structure (interviews with policymakers and experts).

London

There is no single policy framework for migrant integration in the London area. Although the Greater London Authority (GLA) represents a central authority for the greater London area, many competencies related to integration or ‘community cohesion’ remain at the level of boroughs. Between boroughs, there can be significant differences, in terms of relative size of migrant populations, as well as in terms of distribution of specific migrant groups (Gidley and Jayaweera 2010). Hence, this analysis will focus not just on GLA’s policies, but also on policies in two boroughs: Tower Hamlets (an inner-London borough) and Enfield (an outer-London borough).

In spite of its limited formal policy competencies, interviews with local policymakers as well as UK experts show that the GLA, and London’s Mayor in particular, has played an entrepreneurial role in the framing of migrant integration. At various occasions, the former London Mayor Livingstone celebrated the superdiversity of London and dissociated himself for national policy discourses on the restriction of immigration and the importance of Britishness as a concept for integration (interview with UK expert). Furthermore, the GLA has advocated a relatively (when compared to the national level) comprehensive approach to the integration of migrants, in particular, refugees and migrants from Central and Eastern European countries. The ‘London Enriched’ program that the GLA (2009) launched to this aim provides an integration framework that is both more ambitious as more comprehensive than national policies in this area. It develops a comprehensive approach to areas of language, housing, employment, health, community safety, children and community development. Also, it sets a much more positive tone in stating that ‘refugees living in security and dignity, sharing with all Londoners the same chances for a decent quality of life and the opportunity to contribute to London and its development (ibid: 5)’. In an effort to develop a coherent policy framework in which relevant actors are involved, the responsibility for the coordination of the refugees integration strategy was devolved to the GLA in 2006 (Vachelli et al. 2011). In this context, the Mayor leads the London Strategic Migration Partnership, in which various relevant administrative bodies are involved as well as boroughs, NGOs and the UKBA.

This relatively independent and entrepreneurial governance framework of the GLA is also reflected in the very particular dynamics in the framing of migrant integration. As Fig. 2b shows, the need for an immediate implementation of a community cohesion policy, which had emerged on the national agenda in 2001, was not so immediately felt by the GLA. Instead, race relations and in particular the race equality frame remained more prominent, signaling an important element of frame divergence with the national level. Only very gradually, in the second half of the 2000s and after the London bombings, did community cohesion emerge as one of the primary policy frames in London (see Fig. 2b). Furthermore, the GLA emphasis on ‘integration’ is clearly manifest in Fig. 2b as well, which stands out in comparison to the national as well as to most other local cases.

In spite of the entrepreneurial approach of the GLA, migrant integration is primarily a concern of the various London boroughs. One of the most ‘superdiverse’ inner-London boroughs is Tower Hamlets. This borough has a very sizeable community of Muslims from Bangladesh, over one third of the borough population (Butler et al. 2011). As Fig. 2c shows, Tower Hamlets has been very active in the area of community cohesion from the very beginning. This strongly suggests frame alignment around the community cohesion frame. Tower Hamlets was widely recognized as an example for local community cohesion policies as well as for its efforts to promote cross-cultural contact (being awarded the Beacon status and the Pathfinder status in 2003) (Cantle 2005; Tower Hamlets Borough Tower Hamlets Borough Council 2003). An important characteristic of Tower Hamlet’s policies was that it mainstreamed community cohesion policies into its broad community plans that spanned various sectors of local policies. It did not develop a separate policy approach. However, this mainstreaming also seems to hide that, in spite of the shift in discourse, there was much more continuity in policy practices before and after the alignment with the community cohesion frame (interviews with local policymakers). The exemplary status that was awarded to Tower Hamlets was interpreted more as a form of recognition for the experience and expertise that the borough had developed over a longer period already. Apparently, for Tower Hamlets, the community cohesion frame was sufficiently broad and flexible to accommodate local policy practices already in existence, while benefiting from some of the national funds associated to this frame.

Enfield is an outer-London borough that has experienced significant immigration in the period 2001–2006 (Vachelli et al. 2011). In contrast to Tower Hamlets, its immigrant population consists primarily of Caribbean and African black groups (Butler et al. 2011). Enfield too has developed its own policy approach both in the area of race relations, with a powerful Enfield Racial Equality Council (EREC), as well as in the area of community cohesion. However, interviews reveal that it articulated its community cohesion approach as late as 2007, so much later than Tower Hamlets. However, as Fig. 2d shows, Enfield too follows the trend towards a community cohesion rather than race quality frame and little or no attention to an integration frame. So, although perhaps later than in Tower Hamlets, frame alignment around a community cohesion frame has taken place here as well. Echoing the national policy approach, the Enfield approach stresses the focus on building trust between people, increasing a sense of belonging as well as building strong and positive relationships between people. At the same time, it continues most of its race relations approach oriented at preventing discrimination, recognizing diversity and creating similar life opportunities for all (Enfield Borough 2003; Enfield Borough 2007). Enfield stands out in stressing the role of local leadership, in particular the figure of a Cabinet member for community safety and the voluntary sector; Enfield just considers community cohesion not only a local issue but also an issue that has to be targeted by local government (Enfield Borough 2003: 69–70).

Scotland/Glasgow

In Scotland, Glasgow has the most sizeable migrant population, though accounting for only 3 % of the Glasgow population. This involves in particular migrants from Central and Eastern European countries as well as refugees. Precisely because migrant integration as a policy concern is limited to only a small number of cities in Scotland, the Glasgow case is deeply entwined with Scotland’s policies in this area. Furthermore, though immigration and race relations are not devolved, community cohesion does belong to a broader range of issues that was devolved to the Scottish government. This means, for instance, that the national turn towards community cohesion policies after 2001 did not apply directly to Scotland and cities like Glasgow.

One of the factors that influences Scottish and Glasgow’s policies in this area is the ageing population that creates an important economic and demographic need for immigration (interviews with local policymakers). Also, there have been no incidents like the race riots in England, though the murder of a refugee in the area of Sighthill in 2001 did create some concern about race relations in Glasgow. This has created a setting in which Scotland and Glasgow have continued to attribute more attention to race relations and race equality than to community cohesion or integration (see Fig. 2e, f). On the Scottish level, this was exemplified amongst others by the One Scotland campaign, stressing that the Scots were an open people that welcomed migrants and diversity. Within Scotland, there were positive experiences with the reception of migrants within local communities, often recognizing the importance of these migrants for the revitalization of the local economy (McGarrigle 2010; Pacione 2005). Scottish government even developed measures to induce migrants to stay in Scotland for longer periods. This positive tone was also shared by the Scottish Nationalist Party (SNP). Furthermore, it was reflected in programs like the ‘Fresh Talent Initiative’ that allowed foreign students in Scotland to stay to work for a couple of years, hoping that these students would eventually stay for a longer period (interview with a local policymaker). Within Glasgow, this was recognizable in various projects such as the Bridges program aimed at giving migrants a relevant work experience or the close cooperation between the Glasgow administration and local NGOs in the ATLAS partnership to combat discrimination and promote equality (ibid). Only very recently, as exemplified in Fig. 2f, Glasgow witnessed a slight increase of attention to community cohesion.

This local policy setting diverged significantly from the national setting, especially since the rise of the community cohesion and ‘parallel lives’ discourses. These events were interpreted as irrelevant to the Scottish context where cultural diversity would be perceived more positively (interview with a local policymaker). In fact, the juxtaposition of the Scottish approach against the English approach seems to have only reinforced the Scottish discourse of openness and accepting diversity.

At the same time, Glasgow (and Scotland at large) had to face the national competencies in the area of migration management. In this context, there have been many cases of fruitful and sometimes less-constructive interaction between Glasgow/Scotland and the national government. One particularly positive experience in this respect involves an asylum seekers dispersal program. This involves an agreement between the city of Glasgow and UKBA to resettle refugees (primarily from the London area, and primarily families) to Glasgow. Adjacent to this program, Glasgow developed various measures to ‘integrate’ these refugees in Glasgow society, to make sure that once their admission was finalized the migrants would stay in Glasgow. However, there have also been less successful efforts, amongst others by the strategic migration partnership of the Scottish association of local authorities (COSLA), to convince national government of the need for a more flexible immigration policy for Scotland in general and Glasgow in particular (interview local policymaker).

Wrapping up the analysis of the UK case, it is clear that there was no clear top-down policy coordination structure. Although in the aftermath of the Milltown riots in the early 2000s, efforts were made for a more central approach, this was in fact not down in a top-down manner but rather via initiatives such as the Cantle and CIC committees that brought together expertise and stakeholders from the national but most of all from the local level. In terms of vertical relations, also the Home Affairs Department and later the Department of Communities and Local Government and specific programs such as the Beacon and Pathfinder programs that promote horizontal policy learning between local governments around specific policy priorities played a key role in bringing together the national and local levels. Such efforts come close to the ideal type of multi-level governance relations involving a level playing field between stakeholders from different levels. The most recent initiatives around the memorandum ‘Creating the Conditions for Integration’ only reconfirm the absence of a clear national policy framework and effort not to impose a one-size-fits-all approach.

This multi-level governance orientation clearly played a role in frame alignment around a community cohesion frame since the early 2000s but leaving space for significant variation at the local level. In the London cases (eventually even for the GLA), community cohesion became the most prominent frame. In Scotland and Glasgow, the community cohesion frame remained much less relevant, due to devolution and very different local (problem/policy/political) circumstances. Nonetheless, the community cohesion frame gradually emerged in Glasgow in as well over recent years, stressing the importance of more horizontal mechanisms for frame alignment. Finally, particularly the Tower Hamlets case shows that even when frame alignment takes place in local policy discourses, the interpretation of this frame may vary.

Migrant Integration Policies in the Netherlands

The Netherlands is a unitary state, where in many areas, the central government plays a leading and coordinating role. Yet, the multi-level character of migrant integration policies has also become increasingly manifest in the Dutch case, with two of its most diverse cities, Rotterdam and Amsterdam, developing remarkably dissimilar approaches.

National Policies

In contrast to the UK approach that lacked a clear national policy framework, the Netherlands developed a relatively strong national policy framework in the early 1980s. This ethnic minorities policy would become exemplary for a so-called Dutch multicultural model of integration. However, already in the early 1990s, this multicultural frame was replaced by an alternative frame that put much more responsibility to the individual migrant and a lesser role for central state. In response to economic recession of the 1980s, a ‘civic integration’ (in Dutch ‘inburgering’) frame emerged that framed migrant integration as social-economic participation of individual migrants (Entzinger, 2003). In this period, the central coordination role was weakened in favour of a stronger role of the various sector departments, in particular housing, education and social affairs. However, the Home Affairs minister continued to be responsible for the coordination of policy measures between various departments. Furthermore, first attempts were made to align the civic integration approach with urban policies in particular, even for a brief period leading to a special Minister for Urban Policies and Integration.

However, after the turn of the millennium, migrant integration would emerge more prominently than before on the national political agenda. Instigated by the prominent rise of populism in the Dutch political arena, first by the Fortuyn Party and later also the anti-immigrant Freedom Party, the Netherlands was amongst the first countries in Europe to go through an assimilationist turn. This politicization led to a renewed effort to develop a strong national policy framework, led by a very entrepreneurial Minister for Immigration and Integration (Minister Verdonk, from 2003 to 2007). It was this Minister that would take the lead in the formulation of the pre- and post-admission civic integration programs that would constitute the most profound policy change in the 2000s.

This led to the rise of a rival frame around the turn of the millennium. The qualitative policy analysis reveals that the assimilationist turn, though very evident in the arena of public and political debate, was not so much framed in assimilationist terms in actual policies. Interviews and document analysis indicate that in policymaking, the concept of ‘integration’ persevered but changed in meaning (shifting balance from socioeconomic integration to sociocultural integration). This ‘integrationist’ frame involved a more social-cultural definition of migrant integration, stressing the importance of Dutch norms and values in relation to integration, the need for migrants to identify with Dutch culture and society, and often also a concern about levels of immigration that could threaten social cohesion in Dutch society. In the same period, the ‘civic integration’ frame also remained on the agenda. In 2003, new measures were announced to make the civic integration programs mandatory and to introduce pre- as well as post-admission integration tests, which were installed in 2006 and 2007, respectively. These civic integration trajectories involve both socio-economic elements, such as language training and orientation on the labour market, as socio-cultural elements, such as knowledge of life test.

The quantitative mapping of policy framing shows how on the national level the assimilationist turn in the early 2000s at first indeed manifested itself in a rise of attention to ‘integration policy’ (see Fig. 3a). However, gradually, attention to ‘civic integration’ took over, especially since 2003–2004, the year when the concrete policy plans for mandatory civic integration tests were announced. This shows that after the initial rise of the ‘integration’ frame in the context of the fierce politicization of migrant integration in the early 2000s, a gradual shift towards ‘civic integration’ took place when the topic became gradually less politicized and more concrete policy ideas were launched.

Fig. 3
figure 3

Policy frames in Dutch national and local (Rotterdam and Amsterdam) migrant integration policies (annual number of documents on the agenda in the period 2001–2011)

Furthermore, Fig. 3a clearly signals that since about 2010 attention to migrant integration, either in terms of ‘civic integration’ or ‘integrationism’, is declining. The conservative minority government coalition that was installed in 2010, initially supported by the populist Freedom Party, only established a Minister for Migration and Asylum, while for some time delegating the responsibility for integration to the Home Affairs Minister (which was changed back after 2 years). This means that there is no longer a special Minister for integration. The institutional responsibility for migrant integration was mainstreamed to various sector departments, and the very term ‘integration policy’ was increasingly avoided. Migrant integration was increasingly framed as an individual responsibility of the involved migrants themselves. This is also manifested in a reorganization of civic integration policies, where migrants have to organize as well as finance the preparation for pre- and post-entry tests themselves. This retrenchment of government from both integration and civic integration was also continued under the new government coalition that was formed in 2012.

Amsterdam

Amsterdam is one of the Netherlands’ most diverse cities, with a long history of immigration. It is often considered as one of the strongholds of the Dutch multicultural model of integration. Indeed, for a long time, Amsterdam’s minorities policy largely developed well-aligned with national developments (Alexander 2007), based on the idea that ‘minorities should integrate while also maintaining their cultural identity’ (City of Amsterdam, 1982). In the late 1990s, this minorities policy made place for a diversity policy, involving a reframing from ‘group-specific policies’ towards ‘problem-oriented policies’ (see City of Amsterdam, 1999). Many group-specific measures were abandoned, at least in formal policy, for a more generic and individual or citizen-oriented approach (interviews with local policymakers).

Whereas Amsterdam’s policy development until the early 2000s were well aligned with the national level, Fig. 3b shows that the dramatic national policy changes in the early 2000s were reflected much less locally. In spite of the revival of a strong national policy framework, top-down frame alignment did not follow in Amsterdam largely because of specific political, problem and policy factors that were very different in Amsterdam than on the national level. Migrant integration did re-emerge on Amsterdam’s local policy agenda but later than on the national level as the Labour party maintained control over Amsterdam after the 2002 municipal elections. In fact, interviews reveal that local policymakers felt negative effects of the changes in national policy discourse that were influencing local attitudes to migrants as well. Several local policymakers indicated that ‘keeping things together’ at the local level had become increasingly difficult in the early 2000s because of the performative effect of national populist discourse, which signals the effect that decoupling can have on policies in multi-level settings.

The terrorist killing of the filmmaker Van Gogh in Amsterdam November 2004 was a particularly powerful local focus-event that did change Amsterdam’s policy framing. As Fig. 3b shows, integration became a more important policy objective after this tragic event. It also shows that civic integration received much more attention in local policies than integration. This preference for civic integration rather than integration reflects the more pragmatic nature of civic integration policies for coping with newcomers and diversity (interviews with local policymakers). This is also manifest in the key policy document published in the 2000s in response to the 2004 Van Gogh killing, ‘We, the people of Amsterdam’ (City of Amsterdam, 2005). This memorandum made a more explicit issue-connection between immigrant integration and anti-radicalization policies. This new policy initiative seems to reinforce the trend towards a local more intercultural approach that stressed the diverse possibilities for identification with the city of Amsterdam as a means creating social cohesion on the city level, contrasting sharply with emerging policies on the national level in that period. In public debate, Amsterdam’s mayor Cohen’s slogan ‘keeping things together’ became exemplary for this pragmatic approach.

Unlike developments on the national level in the late 2000s, Amsterdam’s attention to both integration policy and civic integration increased further (see Fig. 3b). Whereas national government advocates retrenchment of government intervention in these areas, Amsterdam continues to advocate a more active government intervention, in order to be able to face the immigration-related challenges that the city faces. In this sense, the decoupling that originated in the 2000s due to divergence circumstances between Amsterdam and the national level has persisted.

Rotterdam

Rotterdam is another superdiverse Dutch city but with a very different socioeconomic structure as an industrial working city and a different migration history with relatively more low-skilled labour migrants when compared to Amsterdam. Like Amsterdam, Rotterdam developed its own integration approach, following a very different path than national policies as well as when compared to Amsterdam. Rotterdam had already adopted a policy aimed at integration already at the end of the 1970s, ahead of developments on the national level. Rotterdam’s policies were much more oriented at combating social deprivation and housing problems than at cultural emancipation (interviews with local policymakers). Rotterdam became a key policy entrepreneur in the turn towards an integration policy at the national level as well. During the 1990s, Rotterdam largely continued this integration policy aimed at social-economic participation. Furthermore, already in 1991, Rotterdam began a municipal Project Integration for Newcomers (interviews with local policymakers). Much earlier than would be the case for national policies, Rotterdam made participation in these courses mandatory for migrants that received social benefits. With these programs, policy attention shifted from settled minority groups to the ongoing influx of newcomers to the city. Hence, it is not surprising that, as Fig. 3c shows, not only just integration but also civic integration played a considerable role in local policy framing in Rotterdam in the early 2000s.

The years 2002 and 2003 key events in the local political arena would mark a key turning point in Rotterdam’s integration policies. In these years, Pim Fortuyn made his first moves on the Dutch political stage, as leader of the centre-right populist party ‘Liveable Rotterdam’ that would win the 2002 local elections (Van Ostaijen and Scholten 2014). During these elections, immigrant integration was framed in relation to issues as criminality, Islam, radicalization and decline of social cohesion. In response to this sharp politicization, Rotterdam developed a more assimilationist policy that concentrated in particular on neighbourhoods where migrants are relatively overrepresented (Uitermark 2010). This is clearly manifested in the peak of attention to both integration policy and civic integration after the 2002 elections (see Fig. 3c). One of the policy initiatives involved the introduction of the ‘Rotterdam Code’, or a set of basic rules of social behaviour for everybody in Rotterdam. In addition, Rotterdam intensively lobbied the national government to adopt a law on ‘Special Measures for Urban Areas’, which provided the municipalities with means for preventing the settlement of people from low-income categories or with social security benefits in designated urban areas (Uitermark 2010). This was also called the ‘Rotterdam law’ as it was pushed only by Rotterdam, but also implemented only in Rotterdam.

Unlike the Amsterdam case that was marked by divergence with the national level since the end of the 1990s, the Rotterdam case involves elements of convergence with the national level since the turn of the millennium. However, this alignment does not so much to have been a top-down alignment, but rather alignment due to spin-offs from Rotterdam to the national level. Rotterdam was not just a key policy entrepreneur in changing national legislation in order to enlarge the scope of action for local integration policies, it was also one of the key entrepreneurs behind the development of a civic integration policy. Furthermore, even ahead of national developments, the Rotterdam case clearly reveals a decrease of attention to both civic integration and integration at the end of the 2000s (see Fig. 3c). In fact, Rotterdam attempts to mainstream migrant integration and avoid speaking of integration policy as a distinct policy domain.

In sum, the Dutch case differs from the UK case in several important respects. As expected, the Dutch approach clearly involved more efforts to develop a central policy coordination framework. This applied not only to the ethnic minorities policies of the 1980s but also to the assimilationist turn in the 2000s that was clearly driven by politicization on the national level. Unlike in the UK where efforts were made to bring together expertise and stakeholders from the local and national level to develop the community cohesion policy, this assimilationist turn was driven by national developments. This seems to have contributed to a decoupling of national and local policies, especially for the city of Amsterdam where local policies were driven by not only a very different political setting but also a different problem setting (more pragmatic in dealing with diversity) and a different policy setting. Only in the brief period preceding the assimilationist turn efforts were made to connect urban policies and integration, more recent developments towards mainstreaming migrant integration also involve some elements of decentralization. However, these diverging factors at both levels and the absence of a ‘multi-level governance framework’ seem to explain why very little alignment took place around the integrationist frame that emerged in the 2000s.

A similarity with the UK case concerns the importance of the local policy setting for how migrant integration is framed. Not only the different migration settings and different policy histories but also differences in political settings between both cities clearly made a difference to the integration policies in these cities. However, in the Dutch case, in the absence of multi-level channels for interaction, this only reinforced the ‘decoupling’ of national and local policies and the absence of vertical nor horizontal frame alignment. In as far as ‘frame alignment’ did take place, especially between Rotterdam and the national level, this was driven by local entrepreneurship and ‘vertical venue shopping’ rather than by multi-level governance or top-down coordination. In fact, this resembles to some extent the (less successful) efforts of the GLA to put an integration approach on the national agenda and by Glasgow to influence national immigration policies.

Conclusions

This article speaks to the debates in migration studies on national models of integration and on the local turn of integration policies. It shows how and why vertical relations between national and local governments in the UK and the Netherlands can matter to the framing of migrant integration policies. It also shows what the consequences can be of the lack of frame alignment between levels, leading to policy decoupling or even to policy conflict or perverse effects of contradictory policy measures being adopted at different levels.

First of all, in none of the countries the top-down coordination structures were found that are theorized in the national models literature to promote frame alignment around a specific national model. In the UK, a central policy coordination framework has been largely absent; a situation that continues with the more recent ‘Creating the Conditions for Integration’ memorandum. In the Netherlands, efforts to re-establish a national framework in response to politicization in the national political arena only contributed to decoupling between national and local policies rather than to frame alignment.

Secondly, different sorts of vertical relations were found to promote frame alignment, although not as predicted by national models. Especially in the UK case, we found various practices that approximate the ideal type of multi-level governance, such as bringing together expertise and stakeholders from the national and local levels not only in several committees that had a strong impact on policy formulation but also in programs such as the Pathfinder and Beacon programs that promoted ‘horizontal’ policy learning between cities around specific priorities. Such multi-level governance relations were very important factors in frame alignment between the national and local levels around the community cohesion frame. Furthermore, in the Dutch case particularly in Rotterdam, but to some extent also in the UK for both the GLA and Glasgow, we found instances of vertical venue shopping, where (modest) frame alignment was triggered by entrepreneurship of local governments in putting their frame on the national agenda.

Thirdly, to put the impact of vertical relations as analysed in this article in perspective, the analysis also shows the key importance of the specificities of the local setting to the framing of local integration policies. With reference to the debate on national models as well as on whether or not there is a specifically local model of integration that applies across cases, the analysis provided in this study shows that local policies can be framed very differently in response to local political, policy and problem factors. This is a finding that is hardly surprising based on local governance literature but does call for a better understanding of the implications of such locally specific policies for relations between policy levels; how to avoid decoupling or policy conflict? In this sense, the flexibility of how the UK community cohesion frame could be interpreted in Tower Hamlets, and more recently, even ‘diffusing’ to Glasgow, is a case in point. It reveals how some degree of frame alignment may occur across very diverse local settings when the frame is developed though multi-level interaction in the first place and remains sufficiently broad to adapt to very different situations.

Finally, the analysis reveals some of the consequence that the absence of vertical relations to promote (some degree of) frame alignment may have. In the Netherlands, the decoupling of national and local policy frames clearly led to contradicting policy messages. For instance, policymakers from Amsterdam in particular indicated that they had to work hard to ‘remedy’ the negative performative effects of national policy discourse on local attitudes towards migrants. A similar discrepancy could be found in the UK between national community cohesion policies and local more positive policies towards diversity not only in the superdiverse boroughs of London but also in the city of Glasgow.