Defining and transforming local migration policies: a conceptual approach backed by evidence from Germany

ABSTRACT What are the factors that drive municipalities to become active in the field of migration and migration-related diversity? Based on qualitative data from 126 German municipalities, this article proposes a new conceptual framework that allows for larger-scale comparisons and enables theory building on policy (in)activity at the local level. Setting the scene, the article discusses various theoretical approaches and empirical findings from the literature by using a heuristic of four categories of factors: institutional framework (e.g. competencies, discretionary spaces, multilevel governance), structural conditions (e.g. urbanity/rurality, socioeconomic conditions); local discourses (i.e. narratives creating a local space of possibilities), and local key actors (e.g. mayors, street-level bureaucrats). Empirically assessing the relative importance of these factors, the article focusses on local integration plans. In the German context, a municipality usually establishes an integration plan to actively control the outcome of integration processes, e.g. by orchestrating local stakeholders or allocating resources. Thus, the development of an integration plan can be used as a proxy for a municipality’s appetite to strategically engage with migration and migration-related diversity. As a result, the article proposes the ‘ISDA framework of local migration policymaking’, which distinguishes between ‘defining factors’ (institutional framework, structural conditions) and ‘transformative factors’ (discourses, actors).


Introduction
Local migration policymaking has received considerable scholarly attention for quite some time Penninx et al. 2004). Local actors not only use discretionary spaces while implementing national and regional legislation (cf. Coleman 2012; Edlins and Larrison 2020;Ellermann 2006;Farris and Holman 2017;Schultz 2020) but also develop a wide range of original policies, influenced by the local context . Most authors agree that, over time, local governments have become 'more self-confident and self-reliant in formulating [their] own immigrant policies' (Schiller 2017, 267; see also Zapata-Barrero, Caponio, and Scholten 2017). More recently, some municipalities also openly oppose national immigration policies, particularly in the field of asylum (Agustín and Jørgensen 2019). Thus, the line between 'immigrant policies' that are often perceived as truly local, and allegedly purely national 'immigration policies' (Hammar 1985) becomes blurry. Yet, despite these profound changes and the growing body of literature on 'policy processes in the area of migration and (migration-related) diversity' (Scholten 2020, 108), a comprehensive framework on the factors that trigger a municipality's policy (in)activity is still outstanding.
Following Ostrom (2005, 28), 'the development and use of a general framework helps to identify the elements (and the relationships among these elements)' that serve as cornerstones for an analysis. In this vein, this article proposes a new conceptual framework of local migration policymaking that allows for larger-scale comparisons and enables theory building on the matter. The framework builds on instructive findings from preceding research. However, many studies analyse local migration policies based on a small number of cases, mainly larger or medium-sized cities (cf., e.g. Bloemraad, de Graauw, and Hamlin 2015;de Graauw, Gleeson, and Bada 2020;de Graauw and Vermeulen 2016;Martínez-Ariño et al. 2019). Smaller municipalities are less frequently analysed, and systematic large-scale comparisons of different types of municipalities are virtually absent. While there are some notable exceptions to this rule (cf., e.g. Farris and Holman 2017;Fisher Williamson 2018;García and Schmalzbauer 2017;Lawlor 2015;Ramakrishnan and Bloemraad 2008;Ramakrishnan and Lewis 2005), they predominantly centre on the United States. Results can thus be expected to be influenced by 'national paradigms' (Thränhardt and Bommes 2010) of migration research. In addition, larger-scale comparisons usually focus on just a few factors, whereas the relative importance of different factors has not been addressed consistently, particularly for the European context.
Thus, as a first step, we will subsume the scattered factors that are mentioned in the literature into four broad categories: institutional framework, local key actors, local discourses and structural conditions. Drawing on qualitative data across 126 German municipalities of very different kinds, we then aim at accounting for the relative importance of different factors. While our empirical basis is broad, we will use findings on a single policy output only, namely the existence of local integration plans (Integrationskonzepte). In Germany, most local integration plans (LIPs) are not just political statements, as it might be the case with the self-declaration to be a 'welcoming city' in the United States (Huang and Liu 2018). Instead, shaped by New Public Management, LIPs are usually drafted to control the outcome of integration processes by orchestrating local stakeholders, developing efficient organisational structures, and allocating resources (Damm 2019;Reichwein 2018). Some Länder (states) provide funding schemes to develop a LIP. Yet, LIPs remain a voluntary task and can thus be used as a proxy to assess a municipality's appetite to strategically engage in migration matters (Filsinger and Filsinger 2019). To better understand why municipalities become active or remain inactive, we will analyse the factors that co-occur with the existence of a LIP and pay particular attention to deviant cases. Thereby, we will develop assumptions on which factors tend to trigger policy (in)activity and whether some of the factors can mitigate or even override others. Concluding, we will suggest the ISDA framework of local migration policymaking that distinguishes between defining factors (institutional framework, structural conditions) and transformative factors (local discourses, key actors).
2. Setting the scene 2.1. Literature review: the factors influencing local migration policies The 'factors that make and unmake migration policies' (Castles 2004) have been theorised for the national level to consist of economic and social interests, the institutional framework, political culture, migrants' rights, the presence of civil society organisations or social movements, and the national welfare system. To assess whether these factors are also relevant for the local level, one needs to consider the particular conditions of municipalities.
Firstly, regarding the institutional framework, it is generally acknowledged that municipalities operate in a multi-level setting (see, e.g. Heinelt and Vetter 2008, for migration policy see Caponio and Jones-Correa 2018;Kühn and Münch 2019). However, the degree of their (in)dependence and the nature of their relationships to higher levels is disputed. Whilst the general relevance of the respective nation state's inner organisation, e.g. the degree of (de)centralisation, is widely accepted (Caponio and Jones-Correa 2018;de Graauw 2014), most authors agree that an active local approach to migration cannot be ascribed to a decentralisation of tasks only, nor to inaction on the part of higher levels of government (Gulasekaram and Ramakrishnan 2013). Some indications on the relationships between different levels of government can be borrowed from neo-institutional theory. It explains how institutions influence actions either by setting incentives or by influencing the identity-formation of actors and shaping the norms and values they adhere to (Hall and Taylor 1996;Peters 2019; for migration policy see Filomeno 2017; Schiller 2019). Apart from generally defining the degree of local autonomy, institutions established at higher levels of governance can also contain different levels of '(structural) discretion' (Dörrenbächer 2018; see also Bloemraad and de Graauw 2012) for policy implementation. The institutional framework thus comprises the degree of (de)centralisation, legal and structural discretion, as well the institutional rules that govern interactions at the local level.
Secondly, studies have highlighted the role of local key actors. It is widely acknowledged that institutions do not determine actions, but are interpreted by actors and can be used strategically to pursue their interests (Jackson 2010). The role of collective and individual actors in local politics has traditionally been analysed by pluralist or power approaches, which look at the interaction between various public and private actors (cf. Borkert and Caponio 2010, 17-21;Jones-Correa and Wong 2015). Empirical examples include employers calling for open migration policies or civil society actors pressing for humanitarian approaches (Castles 2004, 869;Ehrkamp and Nagel 2014;de Graauw, Gleeson, and Bada 2020;de Graauw and Vermeulen 2016;Harden et al. 2015). Other researchers have put forward the 'parties-do-matter hypothesis' (Schmidt 1995) that merges the foci on actors, institutions and ideas. In this vein, it is often assumed that left-leaning governments pursue more active integration policies (de Graauw and Vermeulen 2016;Günther, Kurrek, and Töller 2019), while conservative local governments seem to pursue more restrictive policies (Ramakrishnan and Lewis 2005). However, the influence of local party politics on policy output is contested and depends on contextual factors such as the political system (Hopkins 2018;Manatschal 2013;Martínez-Ariño et al. 2019;Myrberg 2017) and the degree of decentralisation (Armenta 2017;Hepburn 2014). Moreover, conflicts between actors at different levels as well as between actors within the same level of governance can affect local policies (cf. Baumgärtel and Oomen 2019;Castles 2004, 866f;Chauvin and Garcés-Mascareñas 2012;Spencer and Delvino 2019).
Apart from collective actors, individuals are found to play an important role at the local level. While research often focuses on local elites, such as mayors (Boos-Krüger 2005), studies have widely demonstrated the importance of street-level bureaucrats for shaping policy outcomes (e.g. Dahlvik 2017). Moreover, migrants themselves can become agents in migration policymaking and implementation (Harden et al. 2015;Paret and Gleeson 2016;Ramakrishnan and Bloemraad 2008;Wong, García, and Valdivia 2019).
If local actors do not realise the room to manoeuvre of higher-level regulations, this can be characterised as an 'invisible local arena' (Caponio 2010, 172-177), i.e. a passive local stance despite formal legal discretion. This observation guides attention to the interaction between institutional factors and actors' perceptions, which are, in turn, influenced by (local) discourses.
Thirdly, local discourses shape problem perceptions as well as identities and interests by institutionalising common knowledge (Boswell and Hampshire 2017). They provide a space of possibilities that regulates what can be said, who can say it and in whose name (Béland 2009). In this perspective, institutions themselves are discursively constructed, i.e. discursive shifts can lead to shifts in institutional practices. This understanding of discourses as being interwoven with institutions and practices leads to the notion that local discourses are specifically tied to local conditions (Barbehön et al. 2016). Regarding migration policymaking, discourses can affect both, perceptions of the local situation (e.g. on legal discretion, availability of resources) and of migration and migrants (e.g. as an opportunity or a threat). For example, narratives of migrants as 'deserving clients' of public services can promote local policies to accommodate migrants (Fisher Williamson 2018). Similarly, models of citizenship and belongingboth national (Jones-Correa 1998;Koopmans et al. 2005) and local (Chauvin and Garcés-Mascareñas 2012;Hoekstra 2018)influence local policies and practices.
Discourses are 'intrinsically political' (Howarth and Stavrakakis 2009, 4), i.e. actors can draw strategically on narratives to strengthen existing discourses and impose them on others (Jones and McBeth 2010, 334). Consequently, actors are viewed as constituted by discourses, while at the same time drawing purposefully on narratives to pursue their interests, thereby contributing to discursive shifts. Empirical studies on migration have found evidence for both, the influence of discourses as a factor that shapes the perceptions of actors (Ayeb-Karlsson 2020; Glorius 2020; Matos 2017; Spieß 2018; Vollmer and Karakayali 2017), and for individual or collective actors who purposely use certain narratives or frames to influence how others understand a particular issue (Ahad and Banulescu-Bogdan 2019;Crow and Lawlor 2016;Haynes, Merolla, and Ramakrishnan 2016;Schwiertz and Steinhilper 2020). Consequently, discourses can mediate the influence of institutional and structural conditions. They include elements such as actors' perceptions of the local situation (structural and institutional conditions), local narratives on migration/migrants and diversity, as well as local values and beliefs.
Fourthly, structural conditions are often assumed to have a direct impact on local policymaking. When it comes to the size and composition of the migrant population, one assumption is that higher numbers of immigrants lead to more active and restrictive local policies by creating 'demographic pressure' (Wong 2012; see also Walker 2014). Studies indicate that the rate of increasei.e. the change in the share of migrantsseems to influence local migration policies rather than their number (Walker and Leitner 2011). In addition, some findings hint that higher shares of immigrants with voting rights (de Graauw and Vermeulen 2016) as well as an educated and liberal population (Huang and Liu 2018) might trigger welcoming policies. Moreover, the composition of the migrant population seems to play an important role. For instance, racialized relations in many U.S. American municipalities shape current policies and practices (Ehrkamp 2019;Marrow 2011;Matos 2017).
Other structural conditions that have been included (explicitly or implicitly) in a number of studies are the population size and the municipality's level of urbanisation. Size affects, e.g. the resources available to local authorities, including financial resources, but also information and expertise (Dörrenbächer 2018; see also Lipsky 2010). According to the predominant narrative in migration research as well as the literature on local politics, cities are supposed to take a more active stance on migration issues than rural municipalities. This structural difference can be exacerbated by the (lack of) experience with migration, since cities are generally expected to have a longer history of migration and integration policies (Borkert and Caponio 2010, 9-13), although there is evidence that rural areas with less experience in the field of migration can also pursue active local policies (Fisher Williamson 2018). U.S. American literature distinguishes between 'old' and 'new destinations' (Ehrkamp 2019;Marrow 2011Marrow , 2019, which are sometimes equated with urban and rural areas (Bloemraad and de Graauw 2012), or with municipalities in 'the North' or on the East Coast ('old') and 'the South' ('new') (Ehrkamp and Nagel 2014). Local socioeconomic conditions have also been shown to influence local migration policies, with more financial resources leading to a more active stance on migration (Huang and Liu 2018;Marrow 2011;Scholten 2013). Consequently, structural conditions comprise a number of different elements, first and foremost the 'urban-rural' divide assumed by many researchers and policy makers, but also including the size and composition of the local migrant population, historical experiences with migration as well as the resources available to local public administration.
This brief overview of the factors that make and unmake local migration policies cannot be exhaustive in every detail. However, we will illustrate that the proposed categorisation into four groups of factors (institutional framework, key actors, framing, structural conditions) can be used to better conceptualise local migration policymaking, especially when it comes to assessing the interplay of factors that still remains understudied.

Municipalities in Germany
While there are some comparative studies on municipalities in different European states Caponio and Borkert 2010), they mainly focus on larger cities and provide a cursory assessment of influential factors due to the widely differing national structures that need to be taken into account. This points to the need for a systematic analysis of a larger number of municipalities in one country. Due to the high number of refugees received in recent years who were distributed across the country, as well as its highly heterogeneous municipalities in terms of size and structures, Germany constitutes a good case.
In Germany, the term 'municipality' encompasses different entities. On the one hand, it comprises independent cities (kreisfreie Städte or Stadtkreise, which are mainlythough not alwayslarger cities). Independent cities directly function as lower state administrative authorities. On the other hand, the term 'municipality' comprises districts (Landkreise, sometimes also translated as counties), which are regional associations encompassing a number of district municipalities (kreisangehörige Kommunen, mainly smaller or medium-sized towns and villages). The district level and the independent cities represent the NUTS 3 level of the European nomenclature for territorial units, whereas the district municipalities represent the LAU level.
All municipalities are granted the right to decide on local matters autonomously (German Basic Law, Art. 28 II). However, which matters are considered local depends on the respective policy areas. Moreover, municipalities are legally part of the federated states (Länder), i.e. the municipalities' room to manoeuvre differs between the states. The allocation of competences between districts and their respective district municipalities varies greatly and, for migration issues, has not been researched comprehensively so far.
Many policies in the area of migration are determined at the national level (Schammann 2018). Some of these policies, such as the procedures for granting asylum, are fully centralised, i.e. they are directly carried out by federal or state agencies. In other fields such as housing or healthcare for refugees, implementation is delegated to the Länder and, eventually, to the municipalities. Usually, these tasks are mandatory, i.e. the provision of services, their scope and content are determined by federal and state legislation. At first sight, this seems to leave little scope of action for municipalities. However, German migration law is known to leave a wide room to manoeuvre, for instance by lavishly using ambiguous legal concepts (Ellermann 2006;Eule 2014;Schammann 2015). Additionally, in some policy fields, such as issues of social cohesion or the coordination of volunteers, municipalities can decide if and how to engage ('voluntary tasks' or decentralised policies). Many of the latter activities have to be funded by the municipalities' general budgets and are usually mentioned in the LIPs.

Expectations on local integration plans
Based on the structures and position of municipalities in German federalism and taking into account the assumptions from the literature review, we developed four groups of expectations regarding potentially relevant factors that lead to the development of LIPs.
1. We expect that municipalities with higher autonomy (independent cities, districts) are more likely to have developed a LIP than district municipalities. 2. From the literature review, we would expect urban municipalities and those with a relatively larger and/or growing migrant population to have a tendency towards initiating LIPs. The same would apply to municipalities with relatively more financial leeway. 3. We expect that municipalities are more likely to develop a LIP if local discourses underline their freedom to act. Moreover, we assume that LIPs are more likely to be developed by municipalities that view migrants as an important part of the (local) society. 4. Regarding local actors, we especially expect mayors who prioritise the issue of migration as being particularly prone to establishing a LIP. The presence of strong and well-established advocacy organisations, or local enterprises in need of (foreign-born) employees, can also favour the development of a LIP. In addition, literature on local party politics indicates that left-leaning governments should be more likely to develop a LIP, even if this influence is contested and seems to depend on institutional factors.

Methodology
In order to critically discuss the expectations derived from literature, this paper builds on fieldwork from two research projects (for an overview of the 126 municipalities under research see supplemental online material). In the framework of the project, 'Two Worlds Apart? Integration Politics in Urban and Rural Municipalities' (2W), we selected 92 municipalities from twelve German states (excluding the city-states and Saarland) according to their administrative level and their degree of 'rurality/urbanity' and 'economic strength'. Additionally, we took further criteria into account, such as (migrant) population and political leadership (for the definition of criteria see supplemental online material). In total, we selected 23 independent cities, 24 districts and 45 district-municipalities within these districts. In each municipality, we conducted two qualitative phone interviews (45-80 min each) with experts from local public administration and civil society. Interviews focused on local migration policies as well as institutional change and relations between various actors. Between July 2018 and May 2019, we concluded 182 interviews and produced full audio transcripts. The second project, 'Future for Refugees in Rural Regions of Germany' (FFR), focuses on eight rural districts in four German states and 32 district municipalities within these districts. The study follows a multi-disciplinary and multi-methods approach and assesses the perspective of the local public administration, politicians, civil society and the refugees themselves. Regarding local policymaking, we conducted 151 face-to-face interviews (50-180 min each; concluded February 2019-June 2020; full audio-transcripts) with experts from politics, public administration and civil society.
The case selection avoids the 'urban bias' of preceding studies, yet contains some imbalances. Firstly, some of the municipalities depicted in Table 1 as being in a 'less good' situation than the national average were chosen because they were relatively well-off in their respective state. Financially strong municipalities are thus underrepresented in the sample. Secondly, four states were investigated by both projects with six municipalities appearing in both samples. We made sure to account for this imbalance while interpreting the data. We also took the opportunity to validate the quality of the data acquired by different methods of data collection. Thirdly, we selected municipalities only if refugees were present. Apart from displaying a 'refugee bias', our data cannot account for municipal (in)activity that resulted in or aimed at onward migration. Fourthly, our empirical material suffers from a 'rural bias' (23 cities vs 103 rather rural or very rural municipalities). Yet, it is important to bear in mind that 'rural' municipalities are diverse: the majority of interviewees in rural and rather rural areas were based in small or medium-sized towns. Moreover, the inclusion of (rural) districts and their municipalities allowed for a more balanced sample regarding the administrative level which we used as primary selection criteria: 52 * NUTS 3 (23 urban vs 29 rural) vs 74 * LAU.
Guidelines for the interviews in both projects contained similar questions, amongst them about the existence and relevance of a LIP. The data was not merged but the interviews were analysed using the same method, i.e. qualitative content analysis according to Mayring (2014). This method can be categorised as a predominantly qualitative approach that explicitly builds a bridge between qualitative and quantitative approaches (Mayring 2019). For the purpose of this paper, we, firstly, coded if a LIP existed and checked which factors occurred in that case. These factors were either annotated as metadata on each transcript (if they were gathered outside the interview, e.g. population size, degree of rurality) or constituted a code themselves (especially discourses, key actors). As some of the codes (e.g. a specific narrative) only emerged inductively in the process of coding, the respective code was discussed and agreed on by researchers from both projects. This proceeding allowed us to validate our findings and, thus, to confirm some expectations on the existence of LIPs while others were questioned by a number of deviant cases. Subsequently, these deviant cases were analysed in more detail, triangulating our findings with a complementary analysis of media and policy documents.

Factors that make and unmake local integration plans
In line with the expectations on the effect of institutional frameworks presented above, we found that municipalities with a higher level of autonomy were more likely to have their own LIP. In the 2W dataset, 33 out of 92 municipalities had developed a LIP. On the NUTS 3 level, these were 17 out of 23 independent cities (73.9%) and 13 out of 24 districts (54.2%). According to the interviewees from the local administration at this level, more efforts to formalise a local approach were underway. This was especially true for the remaining six independent cities, which were all in the process of drafting a LIP. On the contrary, when it comes to the LAU level, only two out of 45 district municipalities (4.4%) had their own LIP. In the FFR dataset, three out of eight rural districts had developed a LIP whereas none of the 32 district municipalities did so. This absence of a LIP on the LAU level was sometimes justified explicitly by the fact that the district provided a LIP, which comprised all district municipalities: Currently we only have the [LIP] from the district, which is meant to apply for all municipalities. (District municipality, 2W, #54) In one of the two district municipalities with a LIP, the district made funding for the municipality dependent on compiling a LIP. In this particular case, local activity did not derive from local autonomy, but rather from its limitations.
The embeddedness in the multi-level framework of German federalism entails another factor that limits activity, i.e. uncertainty, which increases along with the number of levels involved. For example, the mayor of a smaller community criticised that national laws were interpreted in different ways by the 16 German Länder: [T]hose are federal laws. But the [Länder] minister of the interior issues action guidelines; guidelines is saying too little, instructions, how that has to be interpreted. And apparently, that [interpretation] is not the same in all states. That is difficult to comprehend. You are often told about cases, when in exactly the same case, there was a different decision. (District,FFR,#B_lk_1_047) The results on the influence of structural conditions were less clear-cut. Urbanity might play a role as all independent cities in the 2W project had a LIP at least in the making, whereas only half of the rural districts had (both NUTS 3). Economic prosperity seemed not to determine the presence or absence of a LIP, as we find LIPs also in municipalities with a 'less good' economic situation, which contradicts the expectations. The same is true for political parties: there was no indication showing that the emergence of LIPs was associated rather with progressive or conservative parties. Data regarding the number and increase of the migrant population is also inconsistent, even though low numbers of refugees are sometimes mentioned as a reason for not developing a LIP: One aspect that appeared striking was the frequent reference to historical experiences with migration. This was true for 87 out of 92 municipalities (94.6%) of the 2W-sampleirrespective of population size and administrative status. Only three district municipalities, all with less than 10,000 residents, indicated that their administration had not engaged with migration-related affairs prior to 2015. Two further municipalities did not specify their migration history. Just like in the 2W data, interviewees in all eight districts of the FFR project underlined historical experiences with migration. The types of migration interviewees referred to largely corresponded to the overall experience with migration to the Eastern and Western parts of Germany after World War II. However, history was interpreted differently by local actors. Some interviewees explicitly referred to historical experiences with migration as basis for today's policies, highlighting continuity: [This] wave of refugees was not the first migratory wave. That means, here in [name of city] we have actually always been concerned with migration […] We got many guest workers as a result of being a location of the textile industry then etc. And especially in the 1990s we have received more than five thousand from the former USSR. […] And as a result […] the relevant integration structures have been established. That is the experience we had. Moreover, we had a small accommodation facility for asylum seekers where we had a very active asylum working group already back in the days. (Independent city,2W,#63) Others, however, regretted that their municipality did not learn from historical experiences: And I am of the opinion that we are somewhat up to making the same mistakes as last time, when the wave of refugees came from the Eastern part of the world.
[…] Who we had well ghettoized somewhere here.
[…] I thought that we had learned from that. (District,FFR,#C_lk_2_094) When looking at these findings, at least two questions quickly spring to the mind. Firstly, why did NUTS 3-municipalities, specifically half of the districts, remain passive when it came to LIPs although they faced rather favourable conditions, particularly a high degree of autonomy (i.e. deviance in terms of unexpected inactivity)? Secondly, why did one out of 45 district municipalities in the 2W project develop a LIP despite unfavourable conditions that seemed to restrain the vast majority of municipalities on the LAU level, and without being forced to develop a concept (i.e. deviance in terms of unexpected activity)?
Regarding the first question of why some districts remain reluctant to develop a LIP, local narratives portraying the legal and structural conditions as unfavourable proved to play an important role in both projects. If the institutional context was characterised by legal uncertainty, this hampered an active local approach to migration and diversity. The additional government level for districts and district municipalities led to more uncertainty as compared to independent cities. As a result of the vertical and horizontal complexity, many interviewees expressed a narrative of financial and/or legal constraints.
As I see it, money is always the thing that restrains us most […] regarding the possibilities that we would like to realize. (District,FFR,#C_lk_2_Akt_090) [We work within the] guidelines that are imposed on us. But we just don't go beyond that. (District,FFR,#A_lk_2_024) In addition, a narrative of inevitable onward migration toward urban regions hampered activity, particularly in rural municipalities: [R]efugees say that they do not want to stay in rural areas, but want to go to the large cities. That is partly understandable, partly obstructive to our integration efforts. (District,FFR,#C_lk_2_097) If you could steer it, you would also steer it for the Germans. You cannot prevent people from moving elsewhere. (District,FFR,#D_lk_1_116) Yet, in some municipalities, demographic change was seen as a catalyst for active policies toward (refugee) integration, usually drawing on a narrative of skilled labour shortage: [We] certainly have a shortage of skilled labor here and, of course, it would be desirable if some could […] settle down. And this is actually an important reason [to initiate integration policies]. (District,2W,#65) In the FFR data, most interviewees, particularly in district municipalities, doubted that an own LIP was necessary. They reproduced a narrative of non-responsibility and referred to the reception of refugees as an extraordinary challenge that would not lead to long-term settlement and integration needs.
The question [of developing a LIP] did not arise because basically we are, I'd say, only in charge of housing […]. I think, in big cities, where people arrive in their ten thousands, this is necessary. (District municipality, FFR, #C_ka_1_071) While most interviewees perceived the institutional framework as a constraint or cited it as a reason for not becoming active, some viewed it as a challenge that could be faced. The one of 45 district municipalities in the 2W project that came up with a LIP on its own, felt the need to draft a plan that was less 'urban' than the one from the district: [We]  Consequently, the LIP proposed by the district was not seen as sufficient or helpful. This points to the need to not only assess institutional embeddedness, but also the matching between these institutions and the respective local situation (or its perception by local actors). In this case, the mayor, together with administrative staff, pushed for a LIP. In other municipalities, we found a mismatch between the goals of the administration and the political leadershipwith the administration being in favour of a LIP. These municipalities did not develop a LIP. Thus, a local coalition to initiate a LIP necessarily involves political leadership. Mayors seem to be of crucial significance particularly in rural municipalities, where they perform a leadership role vis-à-vis other local actors, while their role in medium-sized municipalities was less clear (see also Boos-Krüger 2005).
As mentioned above, none of the 32 district municipalities in the FFR dataset developed a LIP. However, we found some alternative local solutions that could be interpreted as a first step toward a formal LIP. For instance, in a small rural municipality, the mayor initiated three public 'future labs', inviting the local community to participate in developing a strategic approach and ideas for specific local policies concerning migration. The minutes of these events served as recommendations and guidelines for political decisions. Again, the mayor initiated the development of a quasi-LIP despite the constraining institutional framework and the municipality's rural setting. As a motivation, the mayor placed the question of migration and refugee reception within a narrative of historical obligations: [We] have been educated with this issue Third Reich and so on [… and were taught] that we have a special responsibility in this world. [Now] we actually had the possibility […] to help. (District municipality,FFR,#C_ka_3_077) Reinforcing this narrative, another interviewee explicitly established a relationship between specific historical incidents in the district and a progressive approach to migration and migration-related diversity nowadays: In the past, prior to Hitler becoming chancellor, [name of district] was the district in Germany with [a huge] share of votes for the NSDAP [National Socialist German Workers' Party]. […] This has probably contributed to the situation that […] the rural population said: Please not again here. (District,2W,#36) In other cases, a narrative of humanitarianism has the power to override the institutionally assigned responsibilities and legal constraints: Theoretically it would be possible to say: 'This does not concern us, go contact the state administrationthey are responsible, have the personnel, receive the money.' We are doing this for reasons of humanity. (District,2W,#60) Summing up, on the way to understanding the relevance of different groups of factors in making and unmaking LIPs, the data from both projects shows that the institutional framework serves as a powerful factor for the existence of a LIP. Structural conditions, in contrast, seem to play a less important role. Local discourses and key actors (namely mayors) seem to be of great importance and can help to understand deviant cases.

Discussion: drafting the ISDA framework of local migration policymaking
Although the brief analysis presented in this paper cannot provide more than a glimpse into the significance of particular factors, it offers the possibility to tentatively conceptualise their relative importance for a strategic and active local approach toward migration and diversity.
In particular, the institutional framework plays a constraining role for municipalities by limiting their competences. However, if there is a mismatch between local problems (or problem perceptions of local actors) and solutions from higher levels, this can trigger local activity (for politicisation cf. Gulasekaram and Ramakrishnan 2013). This is exemplified by one of the district municipalities that developed a LIP.
Regarding the impact of structural conditions, our findings are less clear. One reason is that structural and institutional factors often coincide in German municipalities. For instance, the lower degree of autonomy of the LAU level frequently correlates with higher rurality and lower numbers of migrants. However, there are four tendencies worth noting: Firstly, cities were more likely to have a LIP than rural districts though being on a similar administrative level, which points towards an urban-rural divide stated in the literature (for LIPs cf. Filsinger and Filsinger 2019). Secondly, a municipality's financial resources cannot explain local (in)activity sufficiently, seemingly contradicting widespread assumptions (cf. Huang and Liu 2018;Marrow 2011;Scholten 2013). Thirdly, the flow of population over time seems to be more important than the population size, with depopulation being cited as an incentive for active integration policies by several municipalities. This finding complements the focus of former research on movements of migrants rather than the general population development (Ehrkamp 2019;Fisher Williamson 2018;Walker 2014). Nonetheless, local migration development also influences local policies, with activities being stimulated in particular if migrants are perceived as an important element of local society. Fourthly, size and composition of the migrant population seem to influence migration policy activities, in particular if migrants are perceived as an important element of local society. This aspect relates to the influence of local narratives that ascribe a certain role to migration and migrants. Moreover, local discourses can intensify perceptions of legally or structurally tied hands. Sometimes, however, if positive perceptions of migration merge with optimistic perceptions of the local room to manoeuvre, discourses can constitute a powerful transformer of allegedly constraining legal and structural facts. In addition, (local) historical experience with migration can motivate an active approach to migration.
The influence of key actors varies. In contrast to implementation studies (cf. Dörrenbächer 2018), street-level bureaucrats in our sample were apparently not powerful enough to trigger the development of a LIP, even if they can become important coalition partners for mayors who pursue this goal. The same is true for civil society actors. Our evidence on the relevance of party politics is mixed; local migration policies in general seem to be rather independent of the party-political composition of local governments. This might be due to interactions with the institutional and political framework (cf. Manatschal 2013; Martínez-Ariño et al. 2019), which should be analysed in future studies.
Based on these findings, we understand both the institutional framework and the structural conditions as defining factors of local migration policymaking. They constitute the restraints and resources of the municipality. In order to convert them into policies and praxis, however, discourses and key actors are crucial. Therefore, we conceptualise them as transformative factors. They are transformative in the way that they, firstly, can mitigate the influence of those defining factors that already point into a specific direction. Secondly, they configure the defining factors that are otherwise indifferent or ambiguous. If transformative factors indicate a different policy output and yet do not make a difference, we assume that either the narratives are not part of a locally embedded discourse and/or that the local (coalitions of) actors are not strong enough. This conceptualisation of defining and transformative factors leads to the 'ISDA framework of local migration policymaking' (Table 2). It comprises assumptions about the relative importance of several specific factors and their indication towards municipal (in)activity.

Conclusion
Drawing on data from 126 German municipalities, our findings show a nuanced picture of the factors that influence local migration policymaking. Contributing to the literature on a local turn in migration politics, we find that municipalities are not entirely 'shifting from a passive to an active role' (Myrberg 2017). Virtually all municipalities under research perceive themselves as operating in an uncertain, horizontally and vertically complex institutional environment. This perception grows stronger with every government level that needs to be considered by local stakeholders. Regarding LIPs, especially district municipalities picture themselves as being stuck in a pattern of passivity. In particular, if the feeling of institutional disadvantages is matched by unfavourable structural conditions, e.g. a high degree of rurality, the defining factors point toward a rather inactive local approach to migration issues. If, however, a strong combination of activating discourses and local actors exists, municipalities can overcome the pattern of passivity against all odds. The findings of our research are necessarily limited in several ways. Firstly, we were not able to assess the role of different timeframes and external events. We would thus encourage future research aiming at both cross-sectional analysis and panel studies. Secondly, our data stems from municipalities in Germany. We would embrace comparative research encompassing municipalities in more and less centralised states, with liberal and restrictive national migration policies, different historical experiences with migration and differing structural conditions. Furthermore, the applicability of our proposed framework should be addressed in other (migration) policy areas to assess the boundaries of transformative factors in general. In particular, research on the influence and intertwining of discourses and agency in more and less tightly constrained institutional settings could prove insightful.
Our data relied strongly on interviews with locally important stakeholders. Therefore, further research is needed to answer the question to what extent discourses and their narratives form part of a local dispositif or are strategically used by key actors. Moreover, future research could identify and specify more of the factors influencing local agency. This regards the political and administrative leadershipor lack thereofas well as civil society and migrant agency. The apparent passivity of many municipalities might be related to a bias toward forced migration, since research hints to more active local strategies in attracting qualified workers (Kühn and Münch 2019). This link between activity and migrant groups should be explored further. Moreover, the role of party politics is still vague. Future research should focus on conditions concerning its relevance and potential interactions with the size of the municipality or differences between policy fields (cf. Günther, Kurrek, and Töller 2019).
These research gaps and challenges indicate that we are still far from understanding local migration policymaking comprehensively. The growing body of literature on local migration policy, however, already provides us with fruitful insights. Conceptualising local migration policymaking as the interplay of defining and transformative factors might serve as one piece in this puzzle of multi-perspectivity.