Governance by Proxy: A Comparative Policy Analysis of the Mainstreaming of Immigrant Integration Governance

Abstract

This paper assesses mainstreaming as a governance strategy that potentially addresses the call for a more integral governance response to wicked policy problems. Our comparative qualitative policy analysis of the mainstreaming of immigrant integration governance contributes to the governance and mainstreaming literature by emphasizing the importance of strong horizontal and vertical coordination mechanisms and the distinction between universal and proxy targeting in mainstreaming. Although policy programmes indeed seem to broaden to a universal mainstreamed approach, in practice they often operate by “proxy”. Furthermore, horizontal coordination structures tend to be weak or not in place at all, again obstructing the embedding of immigrant integration as a mainstreamed cross-cutting governance issue.

Keywords:

• mainstreaming
• governance
• comparative
• policy analysis
• immigrant integration

1. Introduction

Mainstreaming offers a polycentric and universal policy instrument for the governance of wicked policy problems that cut across society and cannot be confined to specific policy siloes or subsystems. Literature on the governance of complex or wicked policy problems, such as gender, disability and environmental change, has shown the policy challenges that come with this type of policy issues. On the one hand, studies have shown how complex issues require a cross-sectoral and universal approach rather than the development of a “policy silo”. On the other hand, studies have shown that the development of such a mainstreamed approach often comes with complications, for instance of a political nature or in terms of organizing a cross-sectoral policy approach.

This paper looks into immigrant integration as a wicked policy problem in France, the Netherlands and the United Kingdom and assesses whether mainstreaming is applied. We look at how and why mainstreaming develops (or does not develop) in the governance of this wicked policy problem. Due to continuous migration over time, societies have become increasingly diverse and integration has become less a concern for specific migrant groups only, but more an issue that affects mainstream society. Furthermore, it is a policy area marked by a high degree of volatility and contestation, in how the problem is to be defined as well as on how it is to be resolved. Bringing together both the complexity of the governance networks and the contestation over the problem definition, this study connects different sides of wicked policy problems that are often studied separately (Candel and Biesbroek 2016; Hoornbeek and Peters 2017).

We develop mainstreaming as a conceptual model for an integral polycentric policy approach to immigrant integration, by distinguishing a targeting and coordination dimension in mainstreaming. Building on the governance and mainstreaming literature we will firstly assess whether mainstreaming is indeed applied as a governance approach to the wicked policy problem of immigrant integration. Secondly, we will analyse what policy factor may either contribute to or inhibit mainstreaming by a multiple streams analysis. The key question we address in this paper is whether France, the Netherlands and the United Kingdom responded to the complexities of immigrant integration in the form of a mainstreamed governance approach, and how differences between the countries can be explained.

In the study we find that while a development away from centralized and group-specific immigrant integration policies can be observed across the cases, this does not lead to a unified mainstreaming approach. Due to its contestedness, particularly in France and the Netherlands, immigrant integration is addressed by “proxy” rather than bringing it into the mainstream. While programmes seem to broaden to a universal mainstreamed approach, they operate by “proxy” instead: targeting immigrants under a substitute category as a means to circumvent the intractability around immigrant integration. Furthermore, horizontal coordination structures tend to be weak or not in place at all, again obstructing the embedding of immigrant integration as a mainstreamed cross-cutting policy issue.

2. Theoretical Framework

The theoretical ambition of this article is to contribute to the literature on the governance of wicked policy problems by an in-depth study of mainstreaming in immigrant integration policies. Building on the literature on wicked policy problems and polycentric governance, such as gender, disability and climate, we build a two-dimensional model of mainstreaming with coordination and targeting as its dimensions.

2.1. Coordination and Targeting

Traditional state-centric modes of government fall short in addressing the complex nature of wicked policy problems. Instead, polycentric modes of governance have merged involving complex modes of coordination and interdependencies between multiple actors (across different policy areas, levels and types of actors). The increased attention to polycentricism led to the emergence of complexity theories in public administration, bringing broad networks of governance actors into the analysis (Klijn and Snellen 2009; Room 2011 in Peters 2017). Additionally, “new governance arrangements” (Howlett and Rayner 2007), “integrated” or “joined up governance” (Briassoulis 2005 in Rayner and Howlett 2009; Vince 2015; Biesbroek and Candel 2016; Tosun and Lang 2017; Trein et al. 2018) and “whole of government” approaches (6 et al. 1999; Christensen and Laegreid 2007) came up, calling for a coordinated approach to cross-cutting policy issues. Literature on multi-level governance (Hooghe and Marks 2001) and horizontal coordination approaches (Peters 2015), furthermore, zoom in on the workings of these coordinated approaches vertically and horizontally.

Although the literature on governance focuses on the coordination of a multiplicity of (networks of) actors, the literature on wicked policy problems also explicitly addresses contestation over policy problems. What is to be defined as a policy problem in the first place? Who is entitled to do so and why does a certain problem definition prevail over another definition? As the complexity of policy problems and the networks around them increased, an understanding over the multiplicity of problem definitions grew too, drawing attention to the way policy problems are defined or framed (Rochefort and Cobb 1994; Schön and Rein 1994). This speaks to the broader literature in policy sciences following the so-called “argumentative turn” or increased attention for the constructed nature of policies and thereby the increasingly contested nature of the policy problems (see e.g. Fischer and Forester 1993; Roe 1994; Yanow 1996; Hajer 2003). The focus in this study is therefore not only on how policies are coordinated but also how they are defined. Bringing together both the complexity of the governance networks and the contestation over the problem definition, this study connects different sides of wicked policy problems that are often studied separately (Candel and Biesbroek 2016; Hoornbeek and Peters 2017).

2.2. Policy Mainstreaming

We propose that these two sides of policy making come together in the mainstreaming literature. The notion of mainstreaming as a policy strategy emerged in several disciplines, ranging from special education in the 1970s and gender mainstreaming in the 1980s, to more recent discussions on the mainstreaming of environmental and disability policies (Semmel et al. 1979; Madden and Slavin. 1983; Bender et al. 1995; Booth and Bennett 2002; Barnes and Mercer 2005; Verloo 2005; Dalal-Clayton and Bass 2009; Priestley and Roulstone 2009; Nunan et al. 2012; Caglar 2013). Across the different policy fields, mainstreaming broadly refers to an amalgam of efforts to abandon group-specific policies and integrate gender, disability or environmental priorities as an integral part of generic policy domains. Besides the focus on the transition towards a cross-sectional approach, the literature emphasizes the importance of warranting an active engagement with the policy problem at hand, and the need to address these policy issues from an integrated perspective, as a way to address the complex network around an issue like gender or environmental concerns. Mainstreaming is considered a strategy to prevent gender or disability from being side-lined into a concern for specific actors only, but rather to bring it into the mainstream as a concern for all actors and policy fields (Verloo 2005). Gender equality, for example, is rarely treated as a policy domain in itself but is rather addressed as a topic that affects men and women across policy domains.

Firstly, mainstreaming is considered a process of transformation aiming for a more integral, cross-sectional approach to specific policy issues. This entails an integrated approach in terms of both the policy domain and the target groups. The literature on mainstreaming speaks directly to the broader policy literature on polycentric governance on both dimensions. As on both dimensions, coordination and targeting, it brings the policy topic to the mainstream, cutting across policy domains and levels, making, for example, gender equality an integral part of all policy making. Mainstreaming as a governance strategy thus asks how to target and who needs to be targeted by policies.

Mainstreaming thus requires an active effort to create universal awareness of the relevance of the policy topic at hand. Furthermore, the discussion of mainstreaming in the various fields of literature carries a clear reference to issues of governance coordination. This reflects in particular the assumption that mainstreaming involves polycentric forms of governance in which multiple actors are involved. As a process of bringing these priorities into the mainstream, mainstreaming requires a strong and clear coordination of the policy responsibilities across different departments and levels of governance – both horizontally and vertically ‒ to avoid the risk of “becoming everyone’s responsibility, yet nobody’s at the same time” (Caglar 2013, p. 340). In summary, on the two earlier identified dimensions of coordination and targeting, mainstreaming can be defined as a coordinated polycentric and universally targeted governance approach.

2.3. Multiple Streams Analysis of (Non-)Mainstreaming

Besides the empirical focus on whether and how immigrant integration policies are “mainstreamed” in the three selected countries, we will also look at why this is (or is not) taking place. Our expectation is that a wicked policy problem like immigrant integration requires mainstreaming as this polycentric and universal approach can address both the contestation and complexity of immigrant integration by offering an encompassing governance approach. However, policy literature shows that there are many other drivers of policy processes that may inhibit (or promote) mainstreaming. Policy making is not only a matter of finding the most efficient policy strategy, as the literature on wicked policy problems so aptly illustrates. In his famous multiple streams approach, Kingdon (1984) distinguishes three streams or “families of processes”: the problem (which policy problems receive attention), policy (which policy solutions are available to the problem) and political stream (whether policy makers have the motive and opportunity to turn it into a policy).

We will search for drivers or inhibitors of mainstreaming in each of these streams. From the problem stream, we expect that mainstreaming is more likely to be initiated when there is an acknowledgement of the wicked nature of the policy problem, as this calls for an integrated policy response. In contrast, when the complexity of immigrant integration is denied, mainstreaming is unlikely to take place as siloed, and targeted policy solutions will be preferred (Spencer 2011). From the policy stream, we expect that mainstreaming is more likely when there is a structure in place for horizontal and vertical coordination of policies as these structures will enable the implementation of mainstreaming (Verloo 2005; Caglar 2013). Finally, from the political stream, we expect that politicization of immigrant integration (in terms of both salience and polarization, see Van der Brug et al. 2015) hinders the development of a mainstreaming approach as this will call for a simplification of the policy problem and quick fixes (Boswell 2011).

3. Methods

3.1. Immigrant Integration as a Wicked Policy Problem

Immigrant integration can be considered a revelatory case of a wicked policy problem as both the policy definition and the coordination of this policy issue are complex and contested. Immigrant integration policies are any policies concerned with migration-related diversity in society. These policies might entail explicit integration programmes, but can also cover educational policies involved with either migration-related diversity in the curriculum or reaching out to students or citizens with a migrant background, or broader neighbourhood programmes targeting the wider social cohesion and belonging of its citizens. While formerly understood as an issue of immigrants only, contemporary integration policies are increasingly understood as dealing with diversity as an issue for the entire society. This article assesses how these (former) immigrant integration policies are mainstreamed. Although immigrant integration governance is often assumed to cut across different policy fields, it is nevertheless often studied as centralized (Alexander 2007), and very few studies focus on the horizontal coordination of immigrant integration governance (Spencer 2011). Due to its intractability, a study of how immigrant integration can be embedded across policy fields is an important subject of study.

Central in immigrant integration governance is the demarcation of who is in and out, a demarcation between the host society and those that need to integrate, and the discussion of whether immigrant integration is best promoted by generic, colour-blind policies or by policies that separately target specific migrant groups ‒ a discussion between the risk of sustaining or reinforcing inequalities in society when specific problems are not met with targeted policies (see Simon and Piché 2012), and the risk of strengthening ethnic and cultural boundaries in society through the degenerative effect of targeted policies themselves (see Schneider and Ingram 1997), the so-called “dilemma of recognition” (De Zwart 2005). De Zwart distinguishes different policy responses to the dilemma of recognition: accommodation, denial (as described above) and replacement, a compromise between denial and accommodation in which the government (or any other involved actor) constructs its own categories, replacing or broadening the dominant categories ‒ alternatively known as a veiled form of “targeting within universalism” (Skocpol 1991).

As mentioned above, due to continuous migration over time, societies have become increasingly diverse and integration has become less a concern for specific migrant groups only, and more an issue that affects mainstream society (Vertovec 2007; Meissner and Vertovec 2015; Grzymala-Kazlowska and Phillimore 2018). Additionally, in the context of a broader rise of populism in Europe, immigrant integration policies are increasingly problematized as they are seen as benefiting migrants over other citizens, making it harder to justify policies specifically set up for immigrants (Mudde 2013) ‒ also known as the “logic of deservingness” (Schneider and Ingram 1997, 2005).

After the backlash against multiculturalism and the assimilationist turn that occurred throughout Europe in the early 2000s (Joppke and Morawska 2003; Vertovec and Wessendorf 2010), mainstreaming is now referred to as the latest trend in immigrant integration governance (Martinelli 2014; Joppke and Eule 2016; Meissner 2018; Scholten and van Breugel 2018; Westerveen and Adam 2019). In 2004, mainstreaming was included as one of the 11 “European Common Basic Principles on Integration” and, although in different forms and shapes, has been on the policy agenda in many European countries ever since. In this article we will assess what that mainstreaming development entailed and what drives the development to (non-)mainstreaming in immigrant integration governance.

3.2. Case Selection

This paper focuses on immigrant integration governance in France, the Netherlands and the United Kingdom. Though characterized by different patterns of immigration, all three cases can be characterized as countries with a long history of immigration and as exemplary of early (modern) migration to Europe and the early development of policies on the incorporation of immigrants, as well as being subject to the more recent assimilationist turn. Within that set of countries, France, the Netherlands and the United Kingdom form a dissimilar case study design in terms of the likelihood that mainstreaming will occur. The national contexts differ in the problem, policy and political contexts (Kingdon 1984). By a study of the governance of a wicked policy problem in these different contexts we aim to aid a better understanding of how and why wicked policy problems are (not) mainstreamed. Below we will briefly sketch the policy context and history for each of the cases, up to the year 2000.

3.2.1. France

In France no explicit immigrant integration policies exist. The French approach to integration has traditionally been characterized by the so-called Republican Model (Favell 1998). In line with the French Constitution it is prohibited to distinguish French citizens according to race, origins or religion. Instead there is a strong focus on equality, citizenship and the obligation of individual citizens to participate in society. Furthermore, the Republican framework is characterized by striving for cultural unity (Brubaker 1992). The recognition of immigrants or minorities is therefore contested in French policies. Integration policies tend to be indirect, implemented as area-based policies, centring around ‘Politique de la Ville’,¹ targeting priority neighbourhoods for development schemes. The area-based policies are centrally coordinated by Acsé (“Agence nationale pour la cohesion sociale et l’égalité des chances”)² While officially considered neighbourhood policies, the implementation contracts are also targeted at immigrants and integration.

Though de facto immigrant integration policies exist, the “colour-blind” Republican approach has remained a cornerstone of the French approach to integration, leading amongst others to a taboo of statistics on ethnicity or race. Rather than a recognition of the complex and contested nature of immigrant integration, the very subject of integration is considered a taboo in France and due to the absence of explicit policies and statistical data is largely invisible on the policy agenda (Amiraux and Simon 2006). Due to its implicit nature of immigrant integration policies, French integration policies are furthermore characterized by dispersion and fragmentation (Withol de Wenden 2011, p. 88). Overall, French immigrant integration policies are characterized by weak coordination structures.

Given the dominance of the Republican narrative, the position of immigrants in French society is heavily debated (Castles et al. 2013, p. 276), fostered inter alia under the pressure of the extreme right during the mid-1980s when integration was turned into a question of national identity (Geddes and Scholten 2016, p. 61), including debates on what it means to be French, and the questioning of allegiances and loyalty (Withol de Wenden 2011, pp. 77–78). We thus label French immigrant integration policies as highly politicized. With this high politicization, weak coordination structures and taboo on integration we expect that the mainstreaming of immigrant integration is least likely to take place in France.

3.2.2. The Netherlands

The Netherlands was one of the first European countries to develop a centralized national policy strategy aimed at the integration of “ethnic minorities” in the 1980s. However, the Dutch approach changed significantly to a more socio-economic “Integration Policy” in the 1990s and subsequently to a more assimilationist policy after the turn of the millennium. Since the early 2000s Dutch immigrant integration policies are characterized by gradual de-institutionalization.

The problematization of immigrant integration has shifted from a socio-economic to a socio-cultural definition, with a certain level of recognition of the complexity of immigrant integration, though more so at the local than the national level. The centrally coordinated Dutch immigrant integration policies have been typically horizontally embedded in other policy domains since the move to broader “integration policies” in the 1990s. Since the early 2000s integration policies have been decentralized to the local level, with the national government largely withdrawing from the field. The Netherlands thus has the coordination mechanisms that we expect to be aiding the mainstreaming of immigrant integration governance.

The Netherlands is a typical example of the symbolic and restrictive turn in integration (Entzinger 2014). Dissatisfaction with effectiveness of previous immigrant integration policies stirred debates on the conditions of integration and belonging. This period also saw the rise of new populist parties such as Lijst Pim Fortuyn (founded in 2002) and the Partij voor de Vrijheid (founded in 2006), which have been very influential on immigrant integration debates (Scholten 2013). The Netherlands thus scores relatively high on the politicization of immigrant integration. In summary, with relatively strong coordination mechanisms and partial recognition of the complexity of integration, but with rather high political contestedness, we expect that mainstreaming is more likely to take place in the Netherlands than in France.

3.2.3. The United Kingdom

Like France, the United Kingdom does not have an explicit immigrant integration strategy; integration-related policies instead appear in two policy fields. Due to the historical links with the Commonwealth, immigrant integration policies in the United Kingdom initially tended to speak of ethnic minorities rather than migrants, and integration measures were targeted under “race relations” and equality legislation. In the 1990s a separate set of policies were developed to target new groups of immigrants, and these policies still target early arrivals such as recognized refugees. In the early 2000s a community cohesion approach was developed, shifting the focus of the broad integration approach from race relations to the promotion of a shared sense of belonging. These policies explicitly targeted longer residing migrants and their children (Spencer 2011; Geddes and Scholten 2016) and typically developed locally and in a decentralized way.

In terms of the problem dimension, immigrant integration policies in the United Kingdom are broadly framed, from civic integration policies for new arrivals to a broader social cohesion issue. The Commission on Integration and Cohesion played a specific role in emphasizing the “wickedness” of immigrant integration, by arguing against a “one size fits all” approach to immigrant integration in 2007, instead advocating adaptive local policies (Geddes and Scholten 2016). The United Kingdom thus scores relatively high on recognition of the wickedness of the policy problem. With embeddedness in the community cohesion approach the strong local, decentralized development of policies, the United Kingdom contains the horizontal and vertical coordination structures we expect to be important for the development of mainstreaming. Although overall the European trend is towards more restrictive and adaptive integration policies, in the United Kingdom it is more migration policies than integration that are politicized (Carvalho et al. 2015, n.p.). While new immigration is heavily debated, there is relative consensus on the integration of the current population. We thus label the United Kingdom as low-level politicization of integration. With the recognition of the wickedness of immigrant integration, the presence of coordination structures and the relatively low level of politicization, we consider the United Kingdom as a most likely case for the mainstreaming of immigrant integration governance.

3.3. Data Collection and Analysis

The data for this project were collected as part of a larger research project (project Upstream).³ The empirical analysis of this paper is based on the data collected as part of this project by scholars in the respective countries and published in different country and comparative reports.⁴ The studies are based on an analysis of policy documents over the period 2000–2014. As mainstreaming speaks to the embedding of immigrant integration into generic policies and policy fields, we have not limited our analysis to immigrant integration policy as an institutional policy domain alone, but have taken a broader view on generic policy areas that focus on immigrant integration. Besides the (former or indirect) policy field of immigrant integration, we have thus also included policies in the field of education and social cohesion in our analysis, as two policy fields closely related to immigrant integration. The policy analysis was complemented with 54 in-depth semi-structured interviews with policy makers, practitioners and stakeholders involved in immigrant integration policy-making and related fields.⁵ Following a fixed template, every country analysis focused on the modes of targeting and coordination of immigrant integration priorities at the national and local level. The country studies are available online (http://www.project-upstream.eu/). This paper is based on a meta-analysis of the project findings.

We mapped and analysed the developments on the targeting and coordination dimension for the 2000–2014 period. In this we identified and assessed the applied methods of targeting and coordination, as depicted in Table 1. Are the policies specifically targeted at immigrants or generically addressing the entire population, and is integration governance centrally or decentrally coordinated between the policy domains of immigrant integration, education and social cohesion on immigrant integration? Secondly, we assessed why mainstreaming had (or had not) taken place by an analysis of the problem, policy and political streams of the respective cases.

Table 1. Schematic representation of mainstreaming on the key dimensions of polycentric governance (coordination and targeting)

Table 2. Schematic representation of cases

4. Findings

4.1. France

French immigrant integration policies are implemented as area-based policies. This consists of the so-called ‘Politique de la Ville’ targeting priorityneighbourhoods for development schemes.⁶ These area-based policies were coordinated by Acsé (“Agence nationale pour la cohesion sociale et l’égalité des chances”)⁷ Originally established⁸ to facilitate the integration of Algerian male workers during colonial times, the organization’s scope was broadened to increasingly larger target groups, including family members and migrant descendants, later also including generic anti-discrimination policies. While originally (indirectly) intertwined with immigrant integration policies, Acsé’s policies are now officially disconnected from any integration policies. In education we see similar area-based forms of targeting. Educational policies are centrally organized by the Ministry of Education, and are linked to other policy areas, particularly via the “Urban Contracts of Social Cohesion” and the neighbourhood approach. Educational policies are now mostly targeted by socio-economic indicators and via area-based policies such as the “Zone d’Éducation Prioritaire”, receiving additional staff and funding. While originally including the number of pupils with a foreign nationality as one of the criteria, this criterion has been decoupled, as we saw for the other neighbourhood policies.

On the other hand, explicit integration policies targeted at migrants in thefirst five years after arrival were introduced. As of 2000 the focus of integration policies shifted to new arrivals, by introduction of the reception and integration contracts (“Contrat d’Accueil et d’Intégration”; CIA), explicitly linking immigration to integration. In 2008–2009 integration policies under this “Programmes Régionaux d’Intégration des Populations Immigrées” (PRIPI)⁹ ‒ aimed at promoting immigrant integration issues cross-sectionally, with inter alia language training, social and professional integration and health ‒ became further demarcated to newly arrived immigrants, for the first five years after their arrival. In education and social cohesion, these targeted measures are aimed at a quick mastering of the French language, and secondly focus on Republican values. After these five years everyone is considered a French citizen alike and former long-term integration programmes are decoupled from the neighbourhood policies.

In addition to the new distinction between short- and long-term integration, a national Ministry was established for “Immigration, Integration and National Identity”. At the same time, following the impetus from the European race directives, a framework to fight (ethnic) discrimination was established and run by the equality body “La haute autorité de lutte contre les discriminations et pour l’égalité” (HALDE). Under the new government in 2012–2013 French integration policies were revised, moving away from the focus on assimilation, and again came to centrally focus on anti-discrimination, focusing on “equality for all”. For this, the “Interministerial Delegate for Integration and Republican Equality” was established: a delegate with a permanent status, though without a dedicated budget, instead relying on the collaboration with the other ministries. To strengthen the universality of the equality priority, the integration and anti-discrimination policies were disconnected from the focus on disadvantaged neighbourhoods, and targeted at all areas instead.

In summary, while always implicit in the French case, policies have been further decoupled from immigrant integration priorities. While the labels were thus further broadened, many policies continued as they were. While long-term integration policies became more implicit, short-term integration policies were highlighted and separated as an explicit target. We thus see a polarizing development in terms of targeting, highlighting short-term targets while dismissing long-term integration targets, reinforcing the French Republican ideal of full adaptation. The distinction we made between specific and universal targeting does not fully cover our findings, given the distinction between short- and long-term targeting. Furthermore, besides the recent effort to install an inter-ministerial coordination mechanism for integration and equality, policies in the field of immigrant integration are fragmented. We thus conclude that we cannot speak of mainstreaming in France.

4.2. The Netherlands

In terms of policy coordination, the Netherlands has significantly decentralized its immigrant integration policy approach. Since 2008 policy responsibilities have been decentralized to the local level with the “Gemeenschappelijke Integratie Agenda”. This collective integration agenda consists of a series of meetings and guidelines to facilitate the decentralization of policy responsibilities from the national to the local level. In the second phase this decentralization coincided with a further horizontal decentralization of immigrant integration governance, as the programme then very explicitly propagated a generic approach and moved away from integration governance as a separate field (“Gemeenschappelijke Integrale Aanpak”, 2012). Immigrant integration policies were further dismantled at the national level in 2011 when the focus shifted to individual responsibility and adaptation to Dutch society, to be coordinated from the local level. The decentralization came to entail an almost complete withdrawal of the national government. As of 2011, the national government is only involved with short-term civic integration programmes, while withdrawing from long-term integration goals, similar to the development observed in France.

Although the decentralization of immigrant integration governance to the local level was guided by the “Collective Integration Agenda”, such a coordination strategy was absent in realizing this envisioned decentralization horizontally across policy sectors and actors. While the former immigrant integration programmes were dissolved into broad, generic programmes no specific strategies were developed to involve associated departments horizontally in immigrant integration issues. Instead the development can be characterized as a one-sided focus on a generalization of immigrant integration governance without coordination of the embedding of immigrant integration priorities in the “mainstream”.

In terms of targeting, the Dutch approach also shows a gradual shift towards a more universalist approach. In several cities a generic citizenship approach emerged in the 2000s (see e.g. Amsterdam and Rotterdam). Rather than speaking of immigrant integration the policies were reframed in generic city citizenship terms. The goal of the generic city citizenship framework was twofold: on the one hand dismissing specific (targeted) policies while striving to achieve migrant diversity, along with integral gender and sexual diversity and anti-discrimination priorities for the entire city. However, here too little effort was made to embed these priorities horizontally; the shift to city citizenship primarily came to entail a dismantling of integration policies. Financial reforms and a broader change in the role of the government further enhanced the withdrawal of local government from (explicit) immigrant integration policies.

This shift comes with the use of implicit targeting, as in the France case. In generic policy fields such as education and neighbourhood policies a development towards indirect targeting is visible, addressing integration issues without framing and targeting these explicitly. While the first immigrant integration policies in the Netherlands developed at the neighbourhood level in Rotterdam, this explicit link between housing and neighbourhood policies and (targeting of) integration was gradually decoupled. While immigrant integration is often discussed as a policy priority in housing and neighbourhood policies to combat segregation and improve the “liveability” in the neighbourhood, it is seldom targeted directly. Dissolving segregation in the neighbourhood remains an undiminished priority throughout the years, but is instead targeted in socio-economic terms. In education a similar trend away from targeted policies and towards targeting by proxy is recognizable, in which the policy priorities shifted from anti-segregation to educational quality alone, no longer decoupling educational policies from integration priorities. Another example of this is the weighing indicator for extra funding in primary education. While previously the country of origin and the level of education of the student’s parents were taken into account to allocate extra funding, in 2006 this was restricted to the parent’s level of education alone.

So, when taking stock of the Dutch case, mainstreaming entailed the withdrawal of national government from the area of immigrant integration, in terms of both coordination and targeting. What stands out here is the focus on vertical decentralization, but lack of horizontal coordination mechanisms. In our model we distinguished centralized and polycentric forms of coordination, as the Dutch case illustrates it is important that the coordination is assured both vertically and horizontally. Moreover, similar to the case of France, we see the use of implicit targeting or proxies, especially needs and area based in the form of education and social cohesion policies. While the Netherlands moved away from a siloed and targeted approach to integration, it can at best only be labelled as incomplete mainstreaming as targeting still happens indirectly and horizontal coordination structures are largely absent. Both of these factors hinder a mainstreamed immigrant integration approach.

4.3. The United Kingdom

Immigrant integration policies in the United Kingdom have typically evolved in a decentralized fashion, particularly at the local level, and are developed under the heading of community cohesion, by the Ministry of Housing, Communities and Local Government (MHCLG). In 2010 the “Big Society” programme was launched as part of the broader localism agenda. This meant a shrinking state and a shift to civil society to deliver services at the local level. Feeding into the existing “community” framework, the Big Society philosophy also coincided with large retrenchment measures. In contrast to earlier policies in the field that developed at the local level, the community cohesion frame was initiated and coordinated from the Home Office. While there is still a strong emphasis on implementation at the local level, the policy is coordinated from the national level. In 2012 the “Creating Conditions for Integration” framework was launched, mostly emphasizing the individual agency and responsibility of the migrants, decentralizing immigrant integration to the level of civil society and local areas. Like France, the United Kingdom also has several area-based programmes, such as the Health and Education Action Zones.

Over the years, however, the United Kingdom immigrant integration policies have developed a more explicit focus on national identity, integration outcomes and the promotion of neighbourhood cohesion in response to different (inter)national incidents, such as the Milltown riots in 2001 and the 2005 terrorist attacks in London. Rising concerns about the risks of alienation led to an explicit rejection of the former focus on multiculturalism, replacing it by a focus on stronger local bonds and values, but also launching a superdiversity framework in 2005. As of 2000, language acquisition came to play a more central role in immigrant integration debates, with a focus on language as an important facilitator of integration, with an emphasis on the duty of migrants and minorities to learn English. Furthermore, national identity and diversity came to play a more central role in education.

This led to a twofold response in terms of targeting. On the one hand, in line with France and the Netherlands, we see that group specific-funding is avoided, and policies are rather targeted at communities as a whole. On the other hand, so-called “dysfunctional” communities were targeted, defined on ethnic and religious grounds. Indirect forms of targeting are also recognizable. In education, for example, the “Ethnic Minority Achievement Grant” was mainstreamed in the wider “Dedicated School Grant”; additionally the “Pupil Premium” was introduced in 2010 as part of the equality strategy, aiming to support the poorest students, targeted by socio-economic indicators (free school meals) rather than the former migration-related indicators (English as additional language). Both are examples of indirect targeting, as we have also seen in France and the Netherlands.

In conclusion, as in the cases of France and the Netherlands, in the United Kingdom a withdrawal of the central government is visible in the “Community Cohesion” and “Creating Conditions for Integration” frameworks. Although coordinated centrally, the policies are locally implemented and focus on the individual responsibility of the migrants and citizens involved, while the (central) government takes a facilitating role. In comparison to France and the Netherlands immigrant integration policies are relatively well embedded (both horizontally and vertically) in the Community Cohesion Framework. Also in targeting, integration is partially mainstreamed in a universal approach to it in the Framework. However, in education and social cohesion policies, too, indirect forms of targeting immigrants are visible, as we have seen in France and the Netherlands. Although universal targeting and polycentric policy coordination are present in the United Kingdom, we cannot speak of full mainstreaming, as part of the targeting shifted to targeting by proxy rather than universal targeting, and the polycentric coordination structures weakened rather than strengthened in the period under research.

5. Analysis

Our analysis shows that immigrant integration governance in France, the Netherlands and the United Kingdom do not qualify as an integrated mainstreaming approaches. Although we saw a clear trend in terms of both the coordination and targeting dimension away from centralized and group-targeted policies, this trend has remained rather fragmented. In fact, we observed that coordination mechanisms weakened (particularly the Netherlands) and targeting happened indirectly via proxies rather than being universal (France, the Netherlands and, to a lesser extent, the United Kingdom).

Our analysis shows that in the case of a clear wicked policy problem, like immigrant integration, there are various factors that inhibit the development of a mainstreamed approach that would do justice to this type of policy problem. While policy responsibilities were decentralized to (lower) levels of government and partner departments (both nationally and locally), we saw that the immigrant integration policy objectives were not sufficiently embedded in their new environment. This applies in particular to horizontal decentralization (e.g. across policy departments). In the Dutch case, for example, the decentralization of integration to the local level was facilitated by a dedicated programme, but when integration policies were further decentralized across policy departments such programmes were absent. Instead efforts mostly focused on dismissing programmes rather than embedding these priorities horizontally. In the United Kingdom immigrant integration policies always developed in a decentralized fashion, at the local level and in the field of community cohesion. However under retrenchment pressure (as in the Netherlands) we see this involvement erode, further decentralizing responsibilities for integration to the individual level and the local community ‒ although compared to the Dutch case more explicit coordination is involved here. When viewed from the generic policy fields such as education we see a dismantling of integration priorities both in the Netherlands and the United Kingdom, with the replacement of former group-specific policies, as will be explained below. French immigrant integration policies were always indirect, mostly implemented via area-based policies. However, as of 2008 a twofold development took place. While integration priorities for new arrivals (up to their first five years in France) were coordinated, all former integration priorities for long-term residents were dissolved into universal policies, targeting all French citizens alike.

Immigrant integration is strongly problematized across the cases; however, we coincidentally see a move away from (explicitly) targeted policies (for long-term integration). A clear development away from group-specific immigrant integration policies is visible, both in (former) integration policies and in generic fields like education and housing. Both the United Kingdom and France have a tradition of indirect, area-based policies. As described above, in France these policies came to focus strictly on new arrivals only. In the United Kingdom several programmes previously targeting pupils with a migration background were replaced by programmes targeted by socio-economic indicators, though largely covering the same group. In the Netherlands we see a similar development of indirect targeting in both housing and education. Following the logic of “deservingness” (Schneider and Ingram 1997, 2005), it is considered not politically opportune to explicitly target immigrant groups; instead the group is targeted indirectly. We call this targeting by proxy: targeting under a substitute, as a means to circumvent the intractability around integration. While policies, particularly in France and the Netherlands, thus might seem universal, it is rather “targeting within universalism”(Skocpol 1991) that we see here, without explicitly targeting immigrants or explicitly bringing immigrant integration into the mainstream. By separately targeting short-term integration and only addressing long-term integration by proxy, France and the Netherlands avoid bringing integration into the mainstream.

While we distinguish targeting and coordination as dimensions of mainstreaming, our findings illustrate the importance of very precisely defining the conditions under which this leads to an integrated mainstreaming approach. The distinction we made between specific and universal targeting does not fully cover our findings, as we found that proxy targeting is applied as an important strategy to circumvent this distinction. Furthermore the split between short- and long-term residents illustrates new demarcations of target groups. We furthermore distinguished between centralized and polycentric forms of coordination; however, as particularly the Dutch case illustrates, it is important that the coordination is assured both vertically and horizontally.

How then can we explain these cases of non- or partial mainstreaming? From a problem perspective, it seems that the absence of data on migrant minorities in France helps to substantiate the colour-blind approach. Due to this absence, the French case seems to lack a full comprehension and recognition of the wickedness or complexity of immigrant integration governance. French integration policies are furthermore strongly centralized and closely related to the national Republican discourse. The strong identity debates in France seem incompatible with a mainstream prioritization of immigrant integration. Therefore, the use of proxies was required to uphold the idea of a clear central approach.

Also in the Netherlands, the development of a mainstream approach remained partial, albeit for different reasons. While in the Netherlands immigrant integration policies have to some extent always been horizontally embedded, the horizontal coordination weakened with the withdrawal of the national state from (most) integration policies. As in France short-term (civic) integration is coordinated centrally but longer-term integration priorities are dismantled and indirectly targeted. This too coincides with a politicization of integration and a focus on identity debates in which a similar “dilemma of recognition” emerges as in France ‒ more focused on substituting former targeted policies by proxies and avoiding having to speak about integration altogether.

Finally, in the United Kingdom a broader problem definition of immigrant integration, embedded in the Community Cohesion Framework, dates back to a broader inequality and race relations approach to integration. Furthermore, as integration was less perceived as a threat to the national identity in the United Kingdom there seems more room to address integration as a mainstream issue, concerning all citizens. Nevertheless, here too policy reforms are also driven by retrenchment, leading to a dismissal of former group-targeted policies, being replaced by proxy policies, as in the Netherlands. However, where the retrenchment and political context in the Netherlands led to very limited forms of mainstreaming, the political context in the United Kingdom seems to leave more room to prioritize an integrated mainstreaming approach.

The mainstreaming literature emphasizes the need for an integral perspective, to actively bring a policy priority into the mainstream as a concern for all actors and policy fields (Verloo 2005; see also Tosun and Lang 2017). As a process of bringing these priorities into “the mainstream” this approach requires a strong and clear coordination of the policy responsibilities across different departments and levels of governance – both horizontally and vertically, to avoid the risk of “becoming everyone’s responsibility, yet nobody’s at the same time” (Caglar 2013, p. 340). It is exactly this prioritization and coordination that is absent in the case of immigrant integration, particularly in the French and Dutch case, and to lesser extent in the United Kingdom. We argue that this is the case primarily due to the intractability of immigrant integration governance in which rather than a recognition of the complexity of today’s immigrant integration, under political pressure mainstream debates on integration are avoided altogether.

6. Conclusion

This paper examined mainstreaming as a governance strategy toward wicked or complex policy problems, taking immigrant integration as a revelatory case study. Our analysis shows that faced with the complexities and contestation of integration governance, policies seem unable to develop a coordinated polycentric approach such as in an ideal-typical mainstreaming approach. This sets this area apart from policies on other wicked policy problems where more progress has been made in terms of mainstreaming, such as in the area of gender.

Rather than an integrated approach, the mainstreaming of immigrant integration governance in France, the Netherlands and the United Kingdom led to further fragmentation, as a clear prioritization of the decentralized approach was lacking. Horizontal coordination structures tend to be weak or not in place at all and policies were increasingly targeted by proxy, only indirectly targeting integration and people of a migrant background. Rather than facing and addressing the complexity of integration governance, the policies we have analysed seem to avoid addressing integration altogether. The politicized identity debates around integration seem incompatible with mainstreaming as there is no room to prioritize integration as an encompassing concern across society and its citizens. This is particularly the case in France and the Netherlands, and to lesser extent in the less politicized case of the United Kingdom.

Our comparative analysis shows that when faced with wicked policy problems, there are various obstacles to developing a mainstreamed approach. Eventhough the conditions for mainstreaming were met to differentdegrees in the three cases (as depicted in table 2), this does notlead to a full mainstreaming approach in any of the cases but rather leads to a fragmented and proxy approach to immigrant integration

Acknowledgements

The authors would like to thank the UPSTREAM project partners, and specifically Xandra Maan, for their contribution to the UPSTREAM project upon which data this article is based. Furthermore the authors would like to thank Margot Hermus and the anonymous reviewers for their helpful and constructive comments.

Additional information

Notes on contributors

Ilona Van Breugel

Ilona van Breugel is a post-doctoral researcher at the department of Public Administration and Sociology at the Erasmus University Rotterdam. Her research focuses on (local) governance of immigrant integration.

Peter Scholten

Peter Scholten is a Professor of Public Administration at the Erasmus University Rotterdam. His research focuses on the governance of migration and migration-related diversity and the relationship between knowledge and power in the field of migration. Since 2014 Peter has been director of the IMISCOE Research Network and head of the Network Office hosted at the Erasmus University Rotterdam..

Notes

1. Later “Zone Urbaine Sensible”, now “Quartiers Prioritaires”.

2. Later “Commissariat Géneral à l’Égalité” (CGEAT).

3. Funded by the European Commission under the AMIF scheme.

4. See in particular: Maan, X, van Breugel, I, Scholten, P, 2015, The politics of mainstreaming – A comparative analysis of migrant integration governance in Europe, UPSTREAM report; Bozec, G, Simon, P, 2014, The politics of mainstreaming, immigrant integration policies: Case study of France, UPSTREAM report; Jensen, O, Gidley, B, 2014, The politics of mainstreaming, immigrant integration policies: Case study of the United Kingdom, UPSTREAM report; Maan, X, van Breugel, I, Scholten, P, 2014, The politics of mainstreaming, immigrant integration policies: Case study of the Netherlands, UPSTREAM report.

5. 20 in France, 18 interviews in the United Kingdom and 16 in the Netherlands.

6. Later “Zone Urbaine Sensible”, now “Quartiers Prioritaires”.

7. Later “Commissariat Géneral à l’Égalité” (CGEAT).

8. Then operating under the name FAS, later FASTIF and FASILD.

9. Later “Programme Départemental d’Intégration” (PDI).