Understanding the diversity of local diversities: an analysis of the (mis)match between policies and diversity configurations in Dutch municipalities

ABSTRACT

This study examines the relation between local diversity configurations and policies in a selection of Dutch city case studies. Firstly, we identify distinct diversity configurations in Dutch cities, based on a quantitative analysis of 388 municipalities. Subsequently, we conduct a qualitative in-depth study of local policies on migration-related diversity in 10 cities representing specific diversity configurations, to assess the fit between the characteristics in some selected cities and the policy responses. Our study substantiates what has been described in the literature on the local turn in migration studies; the ‘local dimension’ of migration-related diversities and policies reveal a plurality of approaches rather than a one-size-fits-all local model. Differences in urban diversity configurations are reflected in urban policy in specific ways. We also found that inequalities and segregation mattered in terms of the problematization of integration and the choice of policy approach. Our analysis reveals a relationship or ‘match’ between specific urban diversities and specific urban policies. Based on these matches, we propose a fourfold typology of urban diversity policies. Thus, the article seeks to contribute to a more differentiated understanding of the multilevel complexity of diversities at the local level.

KEYWORDS:

• Cities
• governance
• migration-related diversity
• diversity policies

1. Introduction

The local turn in migration studies has drawn attention to how migration-related diversities take shape at the local level (Zapata-Barrero, Caponio, and Scholten 2017). This ‘local dimension’ (Caponio and Borkert 2010) of migration-related diversities has turned out to be a plurality of local settings with different forms or configurations of migration-related diversity. Cities vary not only in terms of the relative size of their migrant population, but also in a variety of migrant populations, degree of segregation, history of migration, and social-economic position of migrants. Consequently, some cities are defined as ‘superdiverse cities’ (Vertovec 2007) or ‘majority-minority cities’ (Crul 2016) while others may be considered relatively new migration cities, segregated cities, or may display different local configurations of diversity. In this context, Meissner and Vertovec (2015) call for a more differentiated understanding of the multilevel complexity of diversities at the local level.

In this study, we examine the relation between diversity configurations and policy approaches in a selection of localities in the Netherlands. Is there a relation between the diversity configuration and the sort of policy approach chosen by a city, and if so, what type of relations do we find? We do not consider this a mono-directional or causal relationship; rather we are interested in whether, and if so how and why, specific configurations of diversity and specific policy approaches go together in specific municipal settings (‘elective affinity’). While it is possible to map the diversity configuration of municipalities using statistical data, the identification of local policy approaches requires a qualitative document analysis and interpretation. Therefore, we employ a mixed-method approach to this study. We first identify distinct diversity configurations, or ideal types, in 388 Dutch municipalities by mapping four dimensions of diversity: migration diversity, segregation, equality, and mobility. Then, from the 12 observed configurations we select typical municipalities from the 6 most different types and conduct an in-depth analysis of migration-related diversity and policies there. Finally, we examine how the local policies on migration-related diversity go together with the configurations of diversity in the cities; do we see patterns of policy approaches and diversity configurations going together?

Our study contributes to a deeper empirical and theoretical understanding of (local) diversities by offering a more differentiated understanding of the complexity of population diversities, as Meissner and Vertovec (2015) have phrased it. First, our study focuses on the broad spectrum of different Dutch municipalities. Thereby we address the knowledge gap in previous literature that tends to focus solely on capital and gateway cities (Schmiz et al. 2020), overlooking smaller localities where diversity is also present in often varying ways (e.g. Bloemraad 2013). Second, we take a multidimensional approach, aiming to capture and jointly analyse the four identified dimensions of local diversity: migration-related diversity, segregation, equality, and international mobility. Third, we identify local diversity types, based on this multidimensional analysis. Lastly, this approach allows us to examine the relations between the local diversity types and modes of local diversity policies. This will help us understand the diversity of urban policy approaches in a way that averts methodological localism.

2. A differentiated approach to local diversities and policies

2.1. Local diversity

Various scholars have tried to develop a more systematic understanding of the diversity of urban diversities (Glick-Schiller and Çağlar 2009; Foner, Duyvendak, and Kasinitz 2019; Pisarevskaya, Scholten, and Kaşlı 2022). Glick-Schiller and Çağlar (2009) differentiated cities in terms of their integration opportunities and global economic positioning. This led to a distinction between top-, up-, down and bottom-scale cities. The strength of this approach is that it helps to understand how different cities provide different integration opportunities (and obstacles). However, a limitation is that it focuses primarily on the economic conditions and opportunities of a city, and not so much on the diversity of city population per se. Following Meissner and Vertovec’s (2015) call for a more differentiated understanding of local diversities, Pisarevskaya, Scholten, and Kaşlı (2022) made a cluster analysis of over 300 European local areas along various dimensions (volume, variety and spread of diversity), revealing 5 distinct clusters: superdiverse cities, migrant minority cities, new diversity cities, low-migration cities, and non-diverse cities. While this classification is more inductive, it does not include analysis of the inequalities between migrant and non-migrant population and the degree of international mobility that cities have.

Building on the literature on local diversity, we have defined four dimensions of diversity. First, we distinguish the migration-related diversity characteristics of the resident population (i.a. Meissner and Vertovec 2015; Van der Greft and Droogleever Fortuijn 2017). This means that we will look not only at how many inhabitants of a city have a migration background, but also at the diversity in terms of their countries of birth, nationalities and migration types. Secondly, we will look at the level of segregation as manifested in the form of residential segregation (Andersson 2013; Bolt and Van Kempen 2010). This can give a view on the extent to which a city is a divided city or to what extent there is mixing in a city. Thirdly, we will look at (in)equality, which includes differences in existing economic status and socio-cultural capital (Beech and Bravo-Moreno 2014; Bolt, Özüekren, and Phillips 2010; Foner, Duyvendak, and Kasinitz 2019). This gives a view on the relative socio-economic positioning of people with a migration background versus residents without a migration background. Lastly, we will look at mobility, which points at new arrivals and departures on top of existing diversity and different types of mobility (i.a. Hirschman and Massey 2008; Rocha-Jiménez et al. 2016; Takenaka and Osirim 2010). This gives a better understanding on whether the city has recently experienced significant international fluidity of its population – either through immigration or emigration.

Based on these four dimensions, various configurations of local diversities can be deductively defined. Each dimension can be imagined as a continuum, the opposite ends of which represent the contrasting ideal types, i.e. high diversity versus low diversity. As our understanding of migration-related diversities is four-dimensional, we need to account for all the possible configurations that could theoretically be observed in empirical cases. Four dimensions, with 2 opposing ends each, can combine in 16 possible ways. These are the configurations of local diversity, as depicted in Figure 1. For instance, upper-left corner ‘4’ represents localities with high diversity, low segregation, low mobility, and low inequality; in the opposite diagonal corner we find configuration ‘2’ – those are localities with low diversity, but high segregation, and equally low mobility and inequality. Based on the empirical results, we will distinguish the different types of local diversity along these four dimensions. In theory, localities of different size may fit in different types, although we expect that many larger cities will at least be in one of the categories with higher levels of diversity.

Figure 1. Theoretical configurations of local diversity.

2.2. Local diversity policy

After mapping variation in local diversity configurations, we examine the link between the diversity configurations and local diversity policy. Amongst the various studies on local migration policy, some authors stress the typical local character of diversity policy. In particular, they discuss the pragmatic problem-coping character of local-level policy-making (Caponio and Borkert 2010; Scholten 2020; Bak Jørgensen 2012) including a typical ‘local approach’ across cities. Additionally, they specifically emphasize the commonality in local approaches even across countries. Other authors claim that local governments rather combine pragmatism and different ideological beliefs in their diversity policies, which would result in different local configurations of ‘paradigmatic-pragmatism’ (Schiller 2015). Yet others point at the unique local approaches, as local policy is understood to be determined by its unique local political and policy context, making policy in each local setting different (Barbehön and Münch 2016). Others explained the differences in policy approaches to diversity via political factors, such as leading party orientations and political consensus (Martínez-Ariño et al. 2019). While these local policies are mostly based on studies of capital or gateway cities, they would leave the broader local contexts largely unaddressed (Bloemraad 2013). Although several studies on medium- and small-sized cities have emerged over the last couple of years (Glorius, Burer, and Schneider 2021; Caponio 2020), this has so far not led to systematic research across cities with specific compositions of residents in order to understand whether their experiences with migration-related diversity correspond with specific diversity policies.

To capture the diversity of approaches towards migration-related diversity in various cities, several dimensions can be distinguished from the literature on local diversity policy. The first dimension is institutionalization of diversity policy, or the extent to which diversity policy is concentrated in one institutionalized ‘integration policy’, or whether the policy approach spans across various policy areas (Hernes et al. 2020). Secondly, we distinguish how and at whom policies are targeted. Such policies may specifically target certain migrant or minority populations or take a universalist approach in a superdiverse setting (Schneider and Ingram 1997; De Zwart 2005; Simon and Piché 2012; Crul 2016).

Finally, a third dimension concerns the relative prioritization of specific aspects of migration-related diversity. The literature distinguishes various facets of migration-related diversity that could be prioritized: the political dimension (political representation, participation), the economic dimension (labour market participation, income), the cultural dimension (emancipation, integration, anti-discrimination), and the spatial dimension (desegregation) (Alexander 2003; Penninx et al. 2004). Along these three dimensions, different approaches to diversity policy can be distinguished.

2.3. The (mis) match between policy and diversity configurations

From a purely objectivist perspective, the assumption would be that local diversity policy is tailored to the needs of a specific local diversity configuration. This means that there would be a perfect match between the diversity configuration and the type of policy approach that is adopted (see Figure 2). In the literature on diversity policy, there are various hypotheses on the relation between diversity configurations and policy. For instance, various scholars have argued that in settings characterized by higher levels of diversity and by more variation within diversity (‘superdiversity’), an institutionalized approach around a distinct ‘integration policy’ focused on specific groups does not work (Vertovec 2007; Zapata-Barrero, Caponio, and Scholten 2017; Crul 2016). Instead, in such settings a mainstreamed approach universally oriented at the whole local population and across domains is expected.

Figure 2. Conceptual framework.

In contrast, in settings characterized by lower levels of migration diversity and the presence of distinct minority groups (for historical reasons, such as guest labour migration, or geographical reasons, such as proximity), a more targeted and institutionalized integrationist approach on specific domains is expected. While in a setting characterized by overall low levels of diversity or temporary residence, an absence of policies is expected (Alexander 2003), inter alia due to low intercultural competences (Glorius 2017).

High residential segregation between migrant and native populations can be interlinked with socio-economic inequalities between these groups because residential segregation is influenced by the patterns of housing stock distribution (van Gent and Musterd 2016). Due to labour market segmentation as well as limitations of their legal status, residents with migration background often belong to lower social strata (Vogiazides 2018). With the upward social mobility, immigrants tend to move into wealthier local areas, thus decreasing the level of segregation (Andersen 2016). In the cities, where the cheaper and more expensive housing stocks are spatially segregated, the socio-economic divide between migrant and non-migrant population will likely be observed. In such settings of striking inequality and segregation between migrants and ‘natives’, we can expect a problematization of migrants.

Such problematization could be reflected either in targeted policies on crime prevention in specific migrant neighbourhoods, targeted labour market and education integration measures, gentrification projects, and eventually, efforts to enhance social cohesion. Previous research suggests that the state governance (i.e. welfare systems) influence on levels of ethnic segregation could be supported or undermined by the housing provision and housing structures (Arbaci 2007; Van der Wusten and Musterd 1998).

3. Methods

Our study follows a muti-method three-step approach. Firstly, we determine to what extent the 388 Dutch localities fit within the 16 ideal types defined along the dimensions of migration-related diversity, segregation, inequality, and mobility. We do so by using statistical data of Dutch National Statistics Office (CBS) and a fuzzy-set ideal type analysis. Secondly, from the observed types, we select 10 localities from 6 types, in which we carry out a qualitative analysis of diversity policy approaches in the period 2011–2017.¹ Lastly, we explore the patterns of policies in association with different diversity configurations in order to find out whether specific diversity policy approaches are observed within certain types of localities. We deliberately avoid making a claim regarding causal relationships between the diversity configurations and policy approaches, because policies can shape the diversity configurations in the localities and be influenced by characteristics of local populations. Thus, we see that these two variables as mutually dependent, shaping each other in a non-linear way, and mediated by many other factors that are beyond the scope of this study. Below we will explain these steps in detail.

3.1. Mapping local diversity configurations

For our research, we use the method of ‘fuzzy-set ideal type analysis’, which was introduced by Kvist (2007). This approach is useful for the multidimensional analysis of similarities and differences between the cases by analysing their fit within theoretical categories. Compared to inductive cluster analysis, it is a more deductive method, or set-theoretic method, which is more useful in combination with our subsequent qualitative analysis of local policies.² Moreover, this is also particularly useful when dealing with heterogeneous types and sources of data. We used non-publicly available datasets from CBS for the year 2017, which come from different sources (administrative registries, databases of Dutch Immigration and Naturalization Service, etc.) and relate to various units of analysis (all individuals of the Netherlands, or only households, or only foreign nationals). The fuzzy-set qualitative comparative analysis allowed us to bring this heterogeneous data together and compare 388 Dutch municipalities across the 4 dimensions in a systematic way. Below, we will explain how we operationalized diversity, segregation, inequality and mobility dimensions, while Appendix I provides a detailed summary of all indicators.

• ‘Diversity’ is understood as migration-related diversity of population in terms of ethnicity (based on the country of birth and parents’ country of birth), nationality (people holding citizenship of various countries), and types of migration (distinguishing between EU migrants, refugees, family migrants, students, and non-EU labour migrants). In our statistical estimations, we use the Gini-Simpson Index, which allows to take into account both the relative group size of various ethnicities, nationalities and migration types, as well as absolute number of distinct ethnicities and nationalities residing in each city.

• ‘Segregation’ is understood as a residential segregation between two groups, that is unevenness of distribution of these groups across neighbourhoods.³ We calculated two indexes of segregation: (1) between ethnically Dutch and ethnically non-Dutch; (2) between Dutch nationals and foreign nationals. The latter shows us the extent of segregation of ‘new’ migrants, whereas the former also considers more residence patterns of long-standing communities (incl. second generation).

• ‘Inequality’ is understood as migrant disadvantage in terms of highest completed level of education and household income, assessed on the basis of difference between ethnically Dutch and ethnically non-Dutch. We focused the analysis of educational inequality only on the lower and higher levels of education because this is where the largest gaps are usually observed. Multiple levels of Dutch educational system were grouped together and the difference in percentage was calculated. The income inequality was computed as a difference in standardized yearly disposable income between households with the ethnically Dutch main earner and household with ethnically non-Dutch main earner.

• ‘Mobility’ is understood as share of people that arrived in a municipality from abroad or left the municipality to abroad. This does not include short-term mobility (i.e. tourism), only residence-related international mobility.

3.2. From indicators to ideal types

The values of all indicators per dimension (Appendix I) were then translated into fuzzy membership scores (see Appendix II for full overview), which define the degree of belonging of each locality within the opposing ideal types of each dimension. The qualitative anchors (thresholds) for assigning cases into sets are defined in a standard QCA procedure. The degrees of belonging to sets are determined by the position of each case on the continuum between these three qualitative anchors (fully out, cross-over, fully in). Two members of the research team looked at the value distribution of each underlying variable and searched for gaps between the cases where to assign the cross-over point – the main borderline between a case considered in or out of the set.

Subsequently, we combined the indicators into four dimensions using ‘AND’ and ‘OR’ logical operations. An ideal type of ‘Highly diverse city’ required high values in all the indicators that are part of it: diversity AND variety of ethnicity, diversity AND variety of nationalities, AND diversity of migration types. An ideal type ‘highly segregated’ required a city to exhibit either high segregation between ethnically Dutch and ethnically non-Dutch, OR high segregation between Dutch and foreign citizens. This was done because one type of segregation does not exclude the other ethnic or nationality segregation can be observed simultaneously or separately from each other, and it was important to define the type in a way that allows to pay attention to differently segregated cities. An ideal type ‘highly unequal’ required both high inequality in terms of education and income. An ideal type ‘highly mobile’ consists of only one variable, which fully determined the belonging to this set. These sets have four logical counterparts (‘lowly diverse’, ‘lowly segregated’, ‘lowly unequal cities (more equal)’, and ‘lowly mobile’ cities). The degree of belonging to these opposite sets is computed by subtracting the fuzzy membership scores of the original sets from 1. Together they constitute 16 possible configurations (Table 1).

Table 1. List of possible 16 logical configurations and conceptual sets they are based on.

Number of configuration	CODE of configuration	How supersets of each dimension constitute types We use the logical operation ‘AND’ (*) to compute fuzzy-set scores
1	mDiS	mob * DIVERSE * ineq * SEGREG
2	mdiS	mob * diverse * ineq * SEGREG
3	mdis	mob * diverse * ineq * segreg
4	mDis	mob * DIVERSE * ineq * segreg
5	mDIS	mob * DIVERSE * INEQ * SEGREG
6	mdIS	mob * diverse * INEQ * SEGREG
7	mdIs	mob * diverse * INEQ * segreg
8	mDIs	mob * DIVERSE * INEQ * segreg
9	MDIS	MOB * DIVERSE * INEQ * SEGREG
10	MdIS	MOB * diverse * INEQ * SEGREG
11	MdIs	MOB * diverse * INEQ * segreg
12	MDIs	MOB * DIVERSE * INEQ * segreg
13	MDiS	MOB * DIVERSE * ineq * SEGREG
14	MdiS	MOB * diverse * ineq * SEGREG
15	Mdis	MOB * diverse * ineq * segreg
16	MDis	MOB * DIVERSE * ineq * segreg

Note: *Big letters represent positive sets, i.e. ‘MOB’ is ‘highly mobile’, while small letters represent the opposite (or negated) set ‘mob’ means ‘lowly mobile’.

3.3. Policy analysis

Following the most-different systems approach, we selected⁴ six types of localities where we examine the diversity governance approaches and compare them within and across the types. To this end, the qualitative policy analysis was performed using the official policy documents on migration diversity in the 10 most typical municipalities within each type. In the selected localities, all official policy documents (memoranda, policy plans) on diversity between 2014 and 2021 (covering the coalition periods 2014–2018 and 2018–2022) were collected and subsequently selected via an online city-council information system (ris or ibabs-online) and/or the municipal website. In total 616 policy documents were collected, out of which 31 policy documents were selected as relevant and applicable to the selected time period. Additionally, the coalition agreements (for 2014–2018 and 2018–2022) were collected for each city, to paint a broader picture of the policy approach of the locality. These 29 documents were coded in Atlas.ti for analysis.

The policy documents were coded on three dimensions: the degree of institutionalization of migration-diversity policies, targeting and the policy domain in which migration diversity is addressed (Table 2). As described in the theoretical framework, the first dimension is institutionalization of diversity policy, or the extent to which diversity policy is concentrated in one institutionalized ‘integration policy’ or whether the policy approach spans across various policy areas. Within this dimension, we distinguish three types. Firstly, mainstreamed policies refer to diversity policies that are explicitly brought into mainstream policy domains, for example, a diversity agenda that spans different policy domains (Scholten 2020). Secondly, institutionalized integration policies, on the contrary, refer to concentrated integration policies within one policy domain, such as a designated immigrant integration programme. Finally, non-institutionalized diversity policies refer to the cities where no diversity policies are present at all, and in which diversity is thus not explicitly addressed in the policies.

Table 2. Codes of policy analysis.

Institutionalization	Targeting	Domains
1.a Mainstreaming	2.a Targeting	3.a Political
1.b Integration	2.b Universal	3.b Economic
1.c Non-institutionalized	2.c Non-targeting	3.c Cultural
		3.d Spatial
		3.e Safety
		3.f Social

Secondly, we distinguish targeting of local policy to analyse how citizens are classified and categorized in policy-relevant categories (see Yanow 2003; De Zwart 2005), and who are considered part of, or responsible for, the policy problem or solution (Bacchi 2008). For the targeting dimension, we distinguish between ‘targeting’, ‘universal targeting’, and ‘non-targeting’. ‘Targeting’ refers to specific groups, e.g. refugees, EU mobile workers or longer residing citizens with a migrant background that are targeted for diversity or integration policies, while ‘universal’ targeting refer to policies that address diversity in the city but explicitly address a wider group in society, for example addressing diversity as a concern for the entire city. ‘Non-targeting’ is used in the analysis when migrants are explicitly not, or no longer addressed.

Finally, a third dimension concerns the policy domain in which the diversity policies are addressed. Building on the literature (Alexander 2003; Penninx et al. 2004), we distinguish the political dimension (political representation, participation), the economic dimension (education and income), the cultural dimension (emancipation, integration, anti-discrimination), the spatial dimension (housing and desegregation policies), and additionally the safety (anti-radicalisation policies) and the social dimension (participation programmes).

4. Results

4.1. Mapping diversity configurations

We start with a quantitative mapping of the local diversity configurations. As we mentioned above, the 4 analytical dimensions could be combined into 16 ideal types. However, four types were not empirically observed. We found that no cities in the Netherlands fit the ideal types combining low mobility and high diversity. This is not that surprising, since our data shows that international mobility positively correlates with four out of five variables describing diversity (see Appendix III, Fig. 1). Only the diversity of migration types correlates negatively with international mobility. This is because cities, that initially were not very diverse, experienced high levels of immigration of very specific groups – EU migrants and refugees, for instance.

The observed 12 ideal types are populated by cities to different degrees – there are very large sets with 146 cases (# 3 MDIS), as well as sets with just 3 cities in it (# 12MDIs, #13MDiS). It means that some ideal types are more frequently observed in empirical reality than others. Table 3 demonstrates the variation among the types in terms of variables included in this analysis, while the detailed description of each type is provided in the Appendix IV.

Table 3. Summary statistics of 12 observed types.

For the purposes of in-depth policy analysis, we have decided to focus on four most dissimilar ideal types. Our main interest in this study is to understand whether mobility and diversity levels matter for the local policy approach and whether such approaches would be different in the cities with higher inequality and segregation levels compared to the cities with lower segregation and inequality. As mentioned earlier, diversity and mobility correlate with each other, as well as migrant disadvantage also correlates with residential segregation (Appendix III, Fig 2).

The distinct feature of the fuzzy-set analysis is that our city cases can have varied degrees of membership in each configuration. Some cities are more fitting to the outlined theoretical ideal types than others (see Figure 3). Since our aim is to make an in-depth analysis, we decided to choose the most fitting localities per type, and two control cases. On the basis of all these premises, we selected to compare (see Figure 4):

• Among highly mobile and diverse municipalities, two localities from type 9 – with high segregation and high inequality, and two localities from type 16 – with low segregation and low inequality

• Among highly mobile but lowly diverse municipalities, two localities of type 10 – with high segregation and inequality, and two localities from type 15 – with low segregation and inequality

• As control cases, we selected among non-diverse and non-mobile localities, one from type 6 – with high segregation and inequality, and one from type 3 – with low levels of segregation and inequality.

Figure 3. Cities fit in theoretical configurations.

Figure 4. Selected cities in theoretical sets for comparison.

4.2. In-depth insight into the selected cities

In this section, we look in-depth into the 10 selected localities to illustrate the dimensions of mobility, diversity, segregation and inequality. First, we explore the dynamic of international mobility in the localities that are all considered highly mobile but in different ways. The vast majority of the internationally mobile population are people with a migration background (90% among immigrating and 70% among emigrating). In the non-mobile localities of Hattem and Culemborg, the share of people coming from abroad was very small, and about half of those were ethnically Dutch. Comparing mobility patterns of Dutch and foreign citizens we notice that the big cities such as Rotterdam, The Hague, and Rijswijk, have significantly more Dutch citizens among newly arrived than the rest of the highly mobile localities (between 24% and 37%). Only in localities with low international mobility, the Dutch nationals were the majority of those arriving to the city from abroad. Localities like Dronten, Rijswijk, as well as Rotterdam and The Hague experience higher outward mobility of Dutch nationals than other localities (between 40% and 50% of those leaving to abroad hold Dutch nationality). The net migration in all these 10 localities was positive in 2017, meaning that more people settle there from abroad than depart. The degree of international mobility needs to be put in the context of absolute numbers too. Whereas in highly diverse localities like Rotterdam and The Hague, we are talking about over 40,000 people immigrating and emigrating, in smaller towns and less diverse localities the count goes down to a few thousands and hundreds, and in Vlieland and in Hattem, we are talking about 100 mobile people across international borders. Still, even a hundred could make an impact if the total population is just above 1000 people (like in Vlieland).

In terms of diversities, we see clear differences in the degree of diversity. For instance, Rotterdam and The Hague have a ‘Simpson diversity index’ of over 0.70, which means that the probability of two random persons meeting each other in these cities and coming from different ethnic backgrounds, is over 70%. This compares to 15% or 20% in Hattem and Het Bildt. Here we also see some relation with population size, although this is not a perfect relation (see Table 4). While the largest two cities (Rotterdam and The Hague) also score highest on the Simpson diversity index the significantly smaller cities of Amstelveen and Rijswijk still have a relatively high diversity index. Amstelveen, for example, has a Simpson Diversity Index of 0.66 compared to 0.74 for Rotterdam, while only 1/6th of its size (and 36th municipality of the Netherlands in terms of population size compared to Rotterdam as 2nd). This might be explained by Amstelveens and Rijswijks proximity to respectively Amsterdam and The Hague, which make these cities typical suburbs (WRR 2020). When we look at the smaller cities the link between population size and the Simpson diversity index becomes even less clear. We also see diversity in terms of types of migrants. In Vlieland and Noordwijkerhout, we find predominantly EU mobile citizens among migrants (97% and 82%, respectively). In het Bildt and Dronten, refugees and EU mobile citizens are the most numerous groups. Amstelveen has a particularly high share of non-EU migrants for the reasons of work (21%), and family reunification (30%).

Table 4. Size of local population and migration-related diversity.

In terms of inequalities, Vlieland and Noordwijkerhout do not exhibit migrant disadvantage in terms of educational attainment (the share of migrants with higher education degree is higher than that of Dutch, and Dutch also have lower educational degrees). However, there are rather significant income disparities between ethnically Dutch and ethnically foreign (with people without migration background earning over 6000 euro a year more). Therefore, these localities cannot be regarded as fully equal, however, they are neither fully unequal. Other cities with less pronounced inequalities are Rijswijk and Amstelveen. In these localities, both education and income inequalities between ethnically Dutch and ethnically foreign are modest, however, still present. In Amstelveen, migrants earn about 1500 euros less annually, and in Rijswijk they earn around 3000 euro less. In these localities, people with migration background generally have higher education levels than the ethnically Dutch citizens (about 5% more), and they are also not too overrepresented in lower education levels.

In terms of segregation, it is important to provide a more nuanced explanation of how segregation patterns differ when measured on the basis of ethnic background and nationality. Segregation on ethnicity includes the second generation, while segregation on citizenship shows whether ‘new’ migrants live in the same or different neighbourhoods than Dutch nationals. In 8 out of 10 cities, we can see that segregation by citizenship is higher than segregation by ethnicity, indicating that ‘new’ migrants are more separated from citizens in their places of residence. In Dronten and het Bildt, segregation of foreign nationals is the highest among the selected localities, which could be related to the fact that both recent refugees and EU labour migrants form a large share of the migrant population there. Two cities that follow a reversed pattern are the two biggest cities in our selection – Rotterdam and The Hague. Especially in The Hague, the index of ethnic segregation is much higher than the segregation index by nationality (0.36 vs. 0.24). This probably has to do with substantial resident populations of former guest-workers and their descendants, who also live in distinct and structurally poorer areas of these big cities – i.e. Rotterdam South or The Hague South-West.

4.3. Analysis of local diversity policies

Subsequently, we analysed the local diversity policies in the selected cities. We looked at a number of policy dimensions: institutionalization in a generic or specific approach, targeting, and the prioritization of specific areas. A summary of our findings from the policy analysis is provided in Table 5.

Table 5. Coded policies in the selected cities.

First of all, in terms of institutionalization, we found mainstreamed policies in Amstelveen, Rijswijk, The Hague, and Rotterdam. Amstelveen is the city with the most explicit diversity policies with its Diversity Agenda. The programme covers different policy domains, explicitly advocating for an integrated approach to diversity to form an inclusive city. This stretches from the integration of ‘foreign communities’ to emancipation programmes for LGBTHI’s and poverty reduction programmes. The locality explicitly addresses the different aspects of diversity in terms of inter alia background, gender, sexuality and religion, as well as explicitly addressing the different policy domains this covers, e.g. the different domains of life of its citizens such as school, work, sports, healthcare and the neighbourhood.

Rijswijk has no explicit policies on diversity but does have a broad anti-polarisation approach covering preventive policies in education, work, and integration. Remarkable is that The Bildt like Rijswijk has no explicit diversity policies but does address the diversity and inclusivity of the municipality in their anti-discrimination policy. While in 2014, Rotterdam had an explicitly targeted integration approach, as of 2018 a more mainstreamed approach is chosen. It is explicitly aimed at an inclusive society and a mainstreamed approach to diversity in the city, making this an integrated element stretching from participation projects to parking policies. The Hague is characterized mostly by its explicit integration policies, though it does also have a mainstreamed citizenship approach covering education and anti-racism policies. The city furthermore states to take ‘an active approach to integration’ given the diverse character of the city (Coalitieakkoord, 2014).

Amstelveen, Rotterdam and The Hague are the only three cities with a dedicated integration/diversity approach, with Amstelveen and the 2018 Rotterdam policy as the most mainstreamed policies, and The Hague and Rotterdam 2014 policies as a more explicit integration policy. Rijswijk, as mentioned, only targets integration through anti-polarisation and het Bildt has its anti-discrimination policy. Migration-diversity policies were initially completely absent (non-institutionalized) in Vlieland, Noordwijkerhout, Dronten, Hattem, and Culemborg. After the increase in the refugee influx, all cities did later develop targeted integration policies for refugees, however, these do not seem to be institutionalized as such. Rather, these concern separate programmes or additional programmes focused on shelter, housing and participation, developed in parallel to the exciting policies.

A second dimension we looked at is the extent of targeting in diversity policy. We found that targeted policies occur in all cities and focuses mostly on asylum seekers (after 2015). As remarked above, this mostly concerns targeted policies or programmes on shelter, housing, and participation. In Noordwijkerhout and Dronten, these policies are combined with policies for labour migrants (and students in Dronten). In Culemborg the ‘native’ citizens are also targeted to facilitate meetings with the newcomers. In The Hague and Rijswijk, the concerns of its current residents over the newcomers are addressed as well. In The Hague, for example, both ‘Hagenaars’ and ethnic minorities who turn away from the (diverse) ‘Haagse’ society are targeted. In Amstelveen, The Hague and Rotterdam, policies focus on other citizens with a migrant background too. As remarked above, these cities have broader diversity policies, targeted at, inter alia, the migrant background of other citizens too.

Inclusive universal policies, targeted at the whole locality, can be recognized in Amstelveen, The Hague, and Rotterdam. Amstelveen has a diversity agenda aimed at creating an inclusive city, through a focus on social cohesion and well-being. While Amstelveen is the only city with such a dedicated diversity agenda, The Hague and Rotterdam do highlight the diversity of their city within their coalition agreements and their integration programmes (see also mainstreamed approach).

All other localities also have social agendas, though not targeted at diversity. While targeting, inter alia, social cohesion, inclusivity, equality, poverty reduction, wellbeing, and participation in these cities, those policies were not targeted or related to migration diversity. In Vlieland, for example, a focus on liveability is mentioned, though specifically focusing on the impact of tourism on the island. Noordwijkerhout focuses on accessible and clear communication (in the context of illiteracy), participation, and ‘liveability’ though without mentioning migration diversity or citizens with a migrant background. Culemborg likewise formulates the ambition to be a ‘caring’ city, focusing on the participation of all citizens, without mentioning any specific groups. Dronten and Hattem finally explicitly focus on an inclusive city. Focused on bonding and feeling at home (Hattem) and social cohesion, as captured in the local term ‘noaberschap’ in Dronten. While explicitly focusing on inclusion, no specific citizen groups are targeted in these approaches.

Finally, regarding the third dimension, the prioritization of specific areas, our analysis found that for all selected localities, policies are mostly focused on the social and economic domains. Some cities also have policies focused on the spatial, cultural and safety domain, while no policies in relation to diversity were addressed in the political domain. All localities have social policies, the policies in the social domain focus mostly on participation and social cohesion (see also above). Only Amstelveen, Rijswijk, The Hague, and Rotterdam also have economic policies focusing e.g. on poverty reduction. Spatial policies in relation to diversity are addressed in The Hague and Amstelveen, with neighbourhood approaches in The Hague and policies targeted at facilitating meetings between citizens and feeling at home in the neighbourhood in Amstelveen. In the other cases, only narrow spatial approaches to housing policies for refugees and/or EU-labour migrants (Rijswijk, Rotterdam, Vlieland, Noordwijkerhout, Het Bildt, and Dronten). Safety policies were addressed in Rijswijk, The Hague and Rotterdam, focusing on liveability and anti-polarisation. In line with the explicit diversity policies, cultural policies are addressed in several localities including the integration and later ‘samenleven’ memorandum in Rotterdam, The integration memorandum of The Hague and the diversity agenda of Amstelveen.

5. Analysis

By bringing together the findings from the policy analysis and the analysis of the diversity configurations, we can analyse the relationship between diversity configurations and local policy approaches. Is there a relation between both, and if so, what does this relation involve? Firstly, the analysis revealed that high level of diversity matches with the degree of institutionalization of migration-related diversity policies. Localities with low diversity levels – Vlieland, Noordwijkerhout, Dronten Hattem, and Culemborg – tend to have a non-institutionalized approach. They have generic social policies without any mention of migration-related diversity. Participation, care, social cohesion, and liveability (‘leefbaarheid’) are the central themes in the coalition agreements of those cities. An exception here is Het Bildt, a locality characterized by low diversity, but high inequality and segregation, that did adopt a mainstreamed anti-discrimination policy. However, it should be remarked that, in contrast to the diverse localities, this is limited to the anti-discrimination only, otherwise diversity policies are not institutionalized. On the contrary, all diverse localities (Rotterdam, The Hague, Amstelveen and Rijswijk) do have an institutionalized approach in place. All four cities have policies explicitly dedicated to the issues of diversity and integration, manifested in distinct and institutionalized policies or policy programmes. This pattern was expected, since the greater share of the population with a migration background, and the higher variety of resident’s origins, languages, and cultures create an incentive for the municipal administration to address migration and diversity in their policies and enable the administrative apparatus to deal with these issues. However, it should be remarked that in Rijswijk, the policies are limited to its anti-polarisation approach only, thus entailing a more narrow approach than in the other three mainstreamed cities. So, while the high level of diversity matches with the degree of institutionalization of migration-related diversity policies, Het Bildt and Rijswijk form an exception with their one-issue policies amongst, respectively, the low-diverse and high-diverse cities. While the four diverse cities also form the largest cities in our sample, thus suggesting a relationship with city size, the level of diversity includes a broader variety of cities showing a different pattern than when looking at city size only (see Section 4.2). Secondly, we found that localities with a high level of diversity follow a mainstreamed and universalistic approach in their policies, with the exception of Rijswijk with its targeted anti-polarisation policies. In the localities where diversity became the new normal of the local social landscape, diversity-related issues are addressed in a variety of policy domains. Inclusion of population with a migration background is no longer a mandate of a specific ‘integration department’, but rather being an integral characteristic of the whole local society is accounted for across different departments of local administration. However, as opposed to universalist mainstreaming, a(n) (additional) targeted policy approach for refugees was observed in all cities. In the wake of the increased influx of refugees in Europe and the Netherlands in the years prior 2017, all cities set-up targeted programmes for asylum seekers and refugees (permit holders), regardless of the size of the refugee population (or broader make-up of the migrant population) in the city. The targeted refugee policies have a special position in this as other migrant groups are not as widely recognized and targeted. While our study revealed a distinctive make up with predominantly EU mobile citizens in Vlieland and Noordwijkerhout, refugees and EU mobile citizens in het Bildt and Dronten, and non-EU work migrants and family reunification in Amstelveen, these groups are only mentioned in two cities. Only in Noordwijkerhout and Dronten there are explicit policies on EU mobile citizens. References to the EU mobile citizens are missing in Vlieland, het Bildt and Amstelveen. Additionally, family migrants and non-EU work migrants also go unmentioned in Amstelveen. While a high level of diversity corresponds with a mainstreamed and universalist policy, targeted policies seem less related to the level of diversity. Targeted refugee policies are developed in all cities regardless of the size of the refugee population, while other distinctive migrant groups are only addressed in two out of five cities.

Thirdly, we found that in the localities with higher levels of inequality and segregation migrant integration was problematized and targeted by policies. While all (inequal and segregated) localities saw some degree of mainstreamed policies, there is a distinction in how diversity is perceived. The Hague and Rotterdam adopted integration policies, explicitly problematizing integration, stressing the responsibility of the immigrant to integrate in the city. While Amstelveen, on the other hand, advocates an integrated approach to diversity to form an inclusive city. Rotterdam had an explicitly targeted integration approach in 2014, this changed to a more inclusive mainstreamed approach in 2018. Among the non-diverse localities, the policy focus on segregated and unequal localities was not as apparently distinct from the towns with lower levels of migrants’ segregation and socio-economic disadvantage.

Fourthly, in terms of policy domains there is a clear difference in the number of different policy areas in which migration-related diversity is addressed between highly diverse and lowly diverse cities. In the localities with low diversity, migration-related issues are part of one or two policy domains: social and in some cases also spatial. On the contrary, in the localities with high diversity levels, matters of migration-related diversity are addressed in many more policy domains, including economic, cultural, and safety domains. This is illustrative of the broad, mainstreamed approach in these localities and often also again reflects city size. Moreover, the economic policy domain is only observed in the highly diverse localities. These policies focus on tackling poverty and welfare dependency, which may have to do with the fact that in highly diverse localities, the heterogeneity of the migrant population in socio-economic terms is also high and constitutes a sizable group that attracts policy attention. As for the distinction among the dimensions of segregation and inequality, no clear patterns of policy domains have been observed, with both types of cities focusing (or not) on safety, culture, and economics, in addition to social and special domains. Of course, this could be due to the limited number of cases that we have analysed.

Finally, regarding the distinction between mobile and non-mobile localities in the context of low diversity, we see that only some of the mobile localities prioritize spatial policies, whereas the control cases with low mobility and low diversity – do not. Thus, mobility per se does not seem to explain the link to spatial policies. The size of total population and of absolute number of immigrants in the non-mobile control cases may matter in this regard. In Het Bildt (MdIS), with the population just over 10,000 residents, segregation of some migrants is very small-scale, therefore it is not salient enough to be addressed by local policy. Also, in Vlieland where the local population is just above 1000 people and less than 100 EU mobile citizens among them, spatial policies do not make sense. Especially since this is a place for tourism, where people come for holidays from the mainland of the Netherlands and from nearby countries, such as Germany. In the cases where spatial policies are in place, they reveal only a narrow approach to housing policies for refugees and/or EU-labour migrants (Rijswijk, Rotterdam, Vlieland, Noordwijkerhout, Het Bildt, Dronten), and this rather strengthens our argument on targeting rather than on spatial policies. Only in the localities characterized by high diversity (and mobility), broader spatial approaches are implemented. Though here too, it does not seem to be mobility or segregation that explains this, but rather the style of institutionalization and targeting, with the more elaborate mainstreamed and universalist approaches leading to the broadest diversity approaches, with a spatial approach as part of that.

These main findings from the analysis are summarized in Figure 5.

Figure 5. Findings of policy analysis.

6. Conclusion

This study brings a differentiated perspective on the relation between local diversity configurations and local policy. It offers what Meissner and Vertovec (2015) describe as a ‘multidimensional reconsideration of diversity’. First of all, our study substantiates what has been described in the literature on the local turn in migration studies; the ‘local dimension’ of migration-related diversities and policies reveals a plurality of approaches rather than a one-size-fits-all local model. Building on a unified dataset for nearly all Dutch municipalities, our study empirically captures this plurality along four different dimensions: migration-related diversities, international mobility, inequality, and segregation. We identified several specific configurations that were most prevalent in the Dutch setting and examined how these different configurations correspond with local policy approaches.

Secondly, our analysis finds that differences in local diversity configurations are reflected in local policies in specific ways. We found that localities with relatively high levels of diversity are more likely to choose an explicit and institutionalized approach to diversity that cuts across a variety of policy sectors and combines elements of targeting and non-targeting. This suggests a relation between diversification or ‘superdiversity’ on the one hand and mainstreaming in integration policy on the other. For instance, we only found an explicitly mainstreamed approach in localities with high levels of diversity. This contrasts with localities with lower levels of diversity, where diversity often remains a more implicit part of generic social policies.

We also found that inequalities and segregation mattered in terms of the problematization of integration and the choice of policy approach. Localities with high diversity, more inequality, and higher levels of segregation tend to have an institutionalized approach that is explicitly targeted at integration, whereas more equal and less segregated localities exhibit a more universal governance approach. With regards to the presence of specific migrant groups, we did not find coherent evidence of local policies focusing on those groups. Only in two out of four localities, this was the case, and one of them was more segregated and the other – less so. Moreover, policies for refugees were explicitly present in all 10 cases, regardless of the size of the refugee population. This means that other factors could be at play in explaining why municipalities focus policies on specific groups (such as refugees) while omitting other groups present in the locality.

Our analysis thus reveals a certain degree of relationship or ‘match’ between specific local diversity configurations and local policies. Adding to the literature on local policy, we show that there is no single local policy model for migration-related diversities. Also, we show that there is more to local policy in this field than the size of the city or the socio-economic position of migrants or the cities (Glick-Schiller and Çağlar 2009). Our analysis shows that the level of diversity, as well as the degree of segregation and inequality in a locality in combination with each other seems to matter in specific policy configurations.

We see a match between high levels of diversity and more universalist approaches, whereas cities with lower levels of diversity are more likely to follow a non-institutionalized approach. Furthermore, we see a match between high levels of segregation and inequality and a more targeted approach, in particular in the localities with higher levels of diversity, whereas cities with lower levels of segregation and inequality are less likely to engage in forms of targeting. With regards to localities with lower levels of diversity, this pattern is not so clear and required further investigation in other explanators, i.e. politics and financial resources.

Based on these matches, we can propose a fourfold typology of local policies (see Figure 6). Localities with high diversity and low segregation/inequality seem to follow an approach which we like to describe as mainstreamed universalism; an institutionalized and distinct approach that is not so much targeted at specific groups, but rather at the whole diverse population (sometimes described as diversity policies). Localities with high diversity and high levels of inequality/segregation seem to follow a more ‘integrationist universalism approach’. Such an approach is still comprehensive and institutionalized, but with more targeting elements, specifically for those migrant groups that experience more socio-economic inequality or/and are problematized by policymakers. Thirdly, we see that localities with low diversity and also relatively low segregation/inequality tend to follow an approach that we would describe as ‘non-institutionalized and non-targeting’. This seems to be a reflection of the lack of problem urgency in these cities, and perhaps also lack of institutional capacity or funding. Finally, localities with low diversity but with high segregation/inequality also tend to follow a non-institutionalized approach, but with some aspects of targeting. This often takes the shape of targeting within other generic policy areas, such as targeted measures in housing, education or social protection.

Figure 6. Typology of local policies.

A strength of this proposed typology is that it has been developed inductively based on the systematic analysis of cities of different types. Yet, this typology requires further refinement and testing. An important limitation is that it is developed on the basis of only Dutch localities; it is likely that national policy context shapes the typology in a specific way, as well as many other factors as common Dutch migration history, politics, etc. So, comparative international research of local diversity policy is needed. Furthermore, a more dynamic and long-term approach is required to really understand how and why specific policies seem to match (or not) specific diversity configurations in cities. Moreover, greater numbers of in-depth qualitative case studies are required to test the validity of this typology, in connection with other important factors. Nonetheless, this analysis is an important step towards a more comprehensive and multilevel understanding of the complexity of local policies and its relation to the ways migration-related diversity manifests itself in local settings.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Notes

1 It wasn’t feasible to collect qualitative data on migration and diversity policies from all 388 localities, therefore we could not pursue an approach of quantitative regression analysis, and chose for another configurational analysis of associations following Ragin’s tradition of Qualitative Comparative Analysis (Ragin 2000; Ragin 2009 ).

2 See Cooper and Glaesser (2011) for a detailed comparison of fuzzy-set classifications with fuzzy cluster analysis.

3 ‘buurten’ in Dutch.

4 The rationale for the selection is further explained in Section 4.1.