Digitalisation Labs: A New Arena for Policy Design in German Multilevel Governance

The federal system has long been seen as one of the biggest obstacles to the digital transformation of the German state. With the enactment of the Online Access Act (OZG), a law that obliges all federal levels to offer their administrative services digitally in a joint portal network by the end of 2022, a new arena for multilevel collaboration has developed in Germany; the so-called digitalisation labs. The labs are intended to bring together representatives of all federal levels, external actors and citizens to promote problem-oriented policy design and the development of innovative policy solutions. Following a neo-institutionalist perspective and using the analytical concepts of multilevel governance and problem-solving, this paper investigates how the institutional settings, internal dynamics and actors’ composition influence policy design processes in the labs. The empirical analysis is built on a qualitative case study of two digitalisation labs in the policy field ‘Immigration and Emigration', and based on ten expert interviews as well as an extensive document analysis. The paper concludes that, by promoting problem-solving, the institutional settings as well as the organisational design and actors’ constellations have influenced the policy design process in several ways.


INTRODUCTION
Decision-making and policy design processes have changed significantly in recent decades to meet the demands of modern society and its complications. In an increasingly digital world with ever-expanding complexity of knowledge and institutional infrastructure, governments and public administrations need to improve their policymaking processes by creating digital and innovative solutions for citizens and companies (Ansell and Gash 2007, 543-544;Janssen and Wimmer 2015, 1-2). To make the digital transformation in politics and administration possible, close collaboration is needed both vertically and horizontally between administrative levels (Layne and Lee 2001, 129-134). To date, the German public sector has been dominated by a decentralised and strongly differentiated administrative landscape. This is on account of its structure as a federal-state as well as municipal self-government. Differences are particularly evident in the digitalisation of the public sector. As a result of this heterogeneous IT landscape, collaboration within the administration is made significantly more difficult. Furthermore, the pursuit of autonomy by the Länder and municipalities hinders the introduction and implementation of e-government reforms in the complex German multilevel system (Kubicek and Wind 2004, 48-49;Martini 2017, 17;Wegrich 2020, 3-4).
Particularly in recent years, public policy and public administration research has focused on the impact of new forms of network-based collaboration to foster collaborative policy design and innovation in the public sector. Under the keyword 'collaborative innovation', governmental arrangements have been examined in which public organisations design and implement new and creative solutions to complicated governance challenges, together with non-state actors, in a problem-oriented policy design process (Sørensen and Torfing 2011, 843-844;Wegrich 2019, 12-14). A special form of collaboration that has been appearing more and more frequently since 2010 is the so-called 'policy (innovation) labs' (Williamson 2015, 4). Policy labs work within or for the public sector or a government entity to support the design and implementation of innovative public policies. In doing so, they involve various stakeholders in policy design processes from a creative, design-and user-orientated perspective (Fuller and Lochard 2016, 1-2). They aim at providing governments at all administrative levels with novel methods for policy design and implementation, taking into account new digital tools and the experiences of different stakeholders (Romero-Frías andArroyo-Machado 2018, 1182). So far, there is a lack of academic literature dealing with the internal dynamics of the labs and how they influence policy design processes (Williamson 2015, 4;Olejniczak et al. 2020, 94).
The implementation of the Online Access Act (OZG) adopted in 2017, has created a novel arena for collaborative policy design in Germany. With the new law, the German government seeks to support harmonisation and better collaboration by creating uniform standards across administrative levels. The law obliges the federal administration, the Länder and municipalities to offer their administrative services digitally by the end of 2022 (Mergel 2019;Wegrich 2020). In doing so, the administration is supported by so-called digitalisation labs. The digitalisation labs represent a new multilevel governance arrangement that seeks to increase acceptance for jointly developed solutions across traditionally rather disconnected levels, typically characterised by an executive federalism that distributes policy formulation and implementation vertically, and in which local self-government is a constitutionally protected norm. The labs are organised by consulting firms and bring together public servants from across all federal levels, alongside digitalisation experts and users, to design and digitalise a highly prioritised public service. A unique feature of these labs is that they challenge classical processes and means of bureaucratic decision-making, focussing on a usercentric approach rather than on formal competencies of the administration, by involving various actors who are traditionally uninvolved in the policy design process in executive federalism.
As with policy labs in general, the internal dynamics and processes of digitalisation labs have barely been scientifically investigated. Against this backdrop, the purpose of this paper is to make an initial scientific contribution to the investigation of this novel arena for policy design in the German multilevel governance system. Following a neoinstitutionalist perspective and drawing on the analytical concepts of multilevel governance and problem-solving, this paper seeks to examine how institutional settings, organisational design and actors' constellations influence policy design process in the labs. To answer this question, a case study of two digitalisation labs in the subject area 'Immigration and Emigration' was conducted based on ten expert interviews, combined with extensive document analysis.
The remainder of the article is structured as follows: in the next chapters, the analytical framework and methodological approach of this paper will be presented in more detail. Afterwards, based on the concepts of multilevel governance and problem-solving the paper analyses the settings, strategies, dynamics and processes of the labs. The paper concludes that both the institutional settings and the organisational design of the labs, as well as the constellations of the actors, influences the policy design process in various ways by facilitating problem-solving behaviour in the laboratories.

PROBLEM-SOLVING IN MULTILEVEL GOVERNANCE SETTINGS
The concept of multilevel governance (MLG) emerged in the 1990s in the course of research on decision dynamics in the European Union. In a nutshell, the concept of MLG deals with systems in which governance takes place predominantly with cooperation across different levels. The term covers not only the vertical relationships between various governments and levels, but also the horizontal relationships between actors within a level, as well as the 'sideways processes' (Maggetti and Trein 2019, 356) with independent and non-state actors. Decision-making processes involving MLG are neither completely centralised nor exclusively made on an individual level, but rather with an interplay of inter-and intragovernmental processes and structures (Hooghe and Marks 2003, 234;Benz 2000, 21-22). Due to the increasing complexity of policy making in a globalised world, the original concept of MLG has been further developed, undergoing various transformations. In today's political science research, the concept is no longer applied exclusively in EU research but also in the context of other multilevel settings, such as federal states (Hooghe and Marks 2003;Stein and Turkewitsch 2008;Irepoglu Carreras 2019;Maggetti and Trein 2019).
Even though the term problem-solving was introduced into MLG research early on (Scharpf 1997b;Benz 2000), the concept remained blurred for a long time (Thomann, Trein, and Maggetti 2019, 38-39). In recent years, political scientists have been increasingly concerned with the question of how the two concepts are related and what influence MLG settings have on problem-solving in policy design processes Trein, Thomann, and Maggetti 2019). In contrast to bargaining, which focuses on the individual interests of actors in the decision-making process, problem-solving is originally seen as a decision-making mode among experts that is characterised by its orientation towards a common interest (Scharpf 1988, 260-261;Benz 2000, 26). Following a procedural understanding and the definition of Maggetti and Trein (2019, 357), problem-solving can be defined as a process in which policy makers: (a) Make policies in the sense of 'puzzling' (on society's behalf) as opposed to 'powering' (Heclo 1974); so as to (b) deal with problems that are perceived important for society by organised groups and/or by policy-makers themselves DIGI TALISATION LAB S: A NEW ARENA FOR P OLIC Y DESI GN (Cohen, March, and Olsen 1972); through (c) the cooperative production of a policy output that is expected to be collectively beneficial in making a contribution to solve the policy problem at stake. (Elgström and Jönsson 2000) In a nutshell, problem-solving is a pattern of behaviour in which decision-makers place solutions to collective policy problems above the particular interests of individual groups (Trein, Thomann, and Maggetti 2019, 341). Problem-solving is of particular relevance when decision-makers are confronted with complex and pressing policy problems that are associated with a high degree of risk, uncertainty, technology or consequence (Thomann, Trein, and Maggetti 2019, 39). Following neo-institutionalist theory, drawing on March (1981) and Streeck and Thelen (2005), Maggetti and Trein argue that the governance structure has an influence on decision-making processes and on 'the policy-solving capacity of the multilevel polity' (Maggetti and Trein 2019, 357). The question of why the MLG arrangements were created is not important, rather the authors assume that organisations change after they are created and can thus also be used for other purposes. In order to analyse the influence of institutional settings for problem-solving in multilevel governance, they identify three structural dimensions: political integration, functional differentiation and decentralisation. Political integration refers to the degree to which the decisionmaking power is concentrated in the centre of multilevel polity, while functional differentiation describes 'the process of functional policy specialization characterising a specific multilevel system' (Trein, Thomann, and Maggetti 2019, 345). The structural dimension of decentralisation deals with the freedom of lower levels of government to implement policies according to their own need, whereas they are usually not involved in the actual policy-making process (Trein, Thomann, and Maggetti 2019, 345).
Hooghe and Marks created a basic concept for the distinction of two types of MLG which can be linked to the concept of problem-solving . Type I governance is characterised by a small number of general-purpose jurisdictions with non-intersecting boundaries. The number of jurisdictional levels is limited to a few levels and the framework is designed to be system-wide and long term. Hooghe and Marks refer to federalism as the 'intellectual foundation for Type I governance' (Hooghe and Marks 2003, 236-237). By contrast, Typ II governance is defined by a large number of intersecting, task-specific jurisdictions which are organised in a flexible, problem-oriented structure across many levels. In empirical reality, the two types of governance co-exist and complement each other. Often, functionally differentiated Type II jurisdictions are embedded in a general-purpose Type I governance structure (Hooghe and Marks 2003, 237-240).
While the advantages of MLG settings lie primarily in their scale flexibility, transaction and coordination costs are often higher than in other systems (Hooghe and Marks 2003, 239). They rise with an increasing number of responsibilities and actors involved in decision-making processes (Scharpf 1997a, 70). Institutional arrangements play a decisive role in dealing with these problems. Two different strategies can be distinguished to deal with transaction and coordination problems: reducing the number of necessary participants and limiting the number of interactions between actors. A reduction in the number of actors involved is particularly evident in Type I governance settings, which are characterised by few non-intersecting jurisdictions, thus limiting the need for vertical and horizontal collaboration between jurisdictions. On the other hand, this small number of jurisdictions leads to increased party competition and distributive bargaining. In Type II governance settings, the number of jurisdictions is not limited. However, transaction and coordination costs in decision-making processes can be reduced by problem-specific, functionally more differentiated jurisdictions. These functional differentiation results in a focus on internal interactions rather than external interactions (Hooghe and Marks 2003, 239-240).
Differing effects of the two MLG types can also be observed concerning the objectives of the actors and the way conflicts are dealt with in decision-making processes. In Type I governance, the actors' focus is primarily on representing the interests of particular groups, and the barriers to exit from the decision-making process are relatively high because these are long-term arrangements in which decisions are regularly made on a variety of issues. This bundling of issues in a few jurisdictions results in ideological conflicts and party competition playing a more important role in Type I governance.
Due to the problem and task-specific focus of the jurisdictions in Type II governance settings, ideological issues take a back seat and the actors involved in the decisionmaking process appear as individuals or members of fluid communities rather than as representatives of a particular group. The voluntary nature of the institutional arrangement allows the actors to leave decision-making processes or to dissolve jurisdictions when they are no longer necessary. At the same time, the functional differentiation and smaller scope of Type II governance structures offer the possibility to avoid conflicts in decision-making processes. Because of the focus on individual issues, decision-making processes can be made independent of other overriding problems and controversies (Hooghe and Marks 2003, 240).
Besides the general type of governance, the nature of the 'denationalisation' (Maggetti and Trein 2019, 360) -upwards, downwards or sideways-can influence the problem-solving capacity. Concerning digitalisation labs, the downwards and sideways denationalisation in Type II governance plays a particularly important role. The downwards denationalisation in Type II governance structures describes the delegation to issue-specific sub-national bodies. Their problem-solving capacity results above all from the possibility of developing customised policy solutions without being restricted by the classical boundaries of competencies. However, new problems can also potentially arise, such as jurisdictional wrangling, when this governance arrangement overlaps with other formal authorities. The problemsolving capacity of sideways denationalisation in Type II governance structures results in particular from the sector-specific knowledge that non-state actors contribute to the policy design process (Maggetti and Trein 2019, 360). While it is assumed that such network governance arrangements can contribute to increased output legitimacy through more adequate and problem-focused decision-making, concerns arise in particular with regard to their democratic accountability. This is because these governance arrangements are often characterised by a low level of citizen participation, a lack of visibility and unclear political responsibilities (Papadopoulos 2007, 471). Even though Type II governance arenas are often not directly involved in the legislative process, they can indirectly exert great influence upon it. Since the capacities and expertise of the legislative authorities are limited, decisions and proposals from policy networks are often simply ratified to avoid transaction costs (Papadopoulos 2007, 474).
The paper is based on the assumption that policy processes are influenced by institutional arrangements, but not completely determined by them. Besides the institutional context, the strategies of the different actors as well as the actor's constellations play an important role (Scharpf 1997a;Mayntz and Scharpf 1995;Kiser and Ostrom 2000). Following this neo-institutionalist perspective and the outlined analytical framework of problem-solving, this paper argues that the institutional and organisational settings of the multilevel governance system and the organisational design of the digitalisation labs, as well as the actor's constellations, influence the policy design process by promoting a problem-solving behaviour in the laboratories.

METHODOLOGY AND DATA COLLECTION
The paper is designed as an explorative, qualitative case study. Since no scientific debate on the subject has yet emerged, a qualitative case study allows an in-depth examination of the subject (Blatter, Langer, and Wagemann 2018, 174). As a case, the digitalisation labs in the subject area 'Immigration and Emigration' were selected. The Federal State of Brandenburg and the Federal Foreign Office are jointly responsible for this subject area. They are officially supported by the Federal Ministry of the Interior, Building and Community (BMI) and the state of Bavaria. Informally, the Länder Baden-Wuerttemberg, Schleswig-Holstein and North Rhine-Westphalia have declared their willingness to support the digitalisation of services. A total of 25 services belong to this subject area (21 on immigration and four on emigration) (Federal Ministry of the Interior, Building and Community 2019, 17). The two services; residence permit for the purpose of gainful employment and formal obligation were given particularly high priority and therefore are examined more closely in two digitalisation labs. The labs are organised by two different consultancies and vary in the number of participants; from 11 (formal obligation) to approximately 20 (residence permit). The subject area can be considered a crucial case, as this policy field shows high degrees of cross-sectoral and multilevel policymaking, touching a wide range of ministerial departments across all governmental levels in Germany. Furthermore, it must currently be considered a particularly salient policy issue. As a result, it is to be expected that there might be a higher potential for political and ideological conflicts between actors in this subject area which could complicate collaborative decisionmaking. Therefore, strategies to promote problem-solving behaviour might be of special importance in the policy design process.
The data basis of this paper consists of 10 expert interviews conducted between January and March 2019, as well as an analysis of official documents on the OZG and the digitalisation labs. Participants and organisers of the labs at all administrative levels, as well as consultants, were interviewed based on semi-structured interview guidelines. This multi-actor approach allows an examination of the internal dynamics and processes of the labs from different angles. Additionally, interviews with the consultants and organisers of the labs offer the opportunity to obtain detailed information on the strategies of the labs, which are not provided in official documents.

DIGITALISATION LABS IN GERMAN MULTILEVEL GOVERNANCE SETTINGS
In the following chapters, the institutional framework, internal dynamics and processes of digitalisation labs will be analysed based on the empirical data surrounding digitalisation labs in the subject area 'Immigration and Emigration'. How multilevel governance settings, the organisational design of the labs and the composition of the actors influence the policy design processes in the labs will be discussed. In doing so, special attention will be paid to the identification and analysis of strategies to promote problem-solving.

Institutional Settings
The German multilevel system is characterised by a fragmentation of competencies both vertically between different federal levels, as well as horizontally between departments. Based on the federal constitution, the German federal system consists of two jurisdictional levels, with autonomous decision-making powers in a non-hierarchical relationshipthe federal and the Länder level (Mayntz 1999, 101). The local level is not directly involved in the legislative process but is represented on the legislative stage by the Länder. Even though the Länder are involved in the legislative process through the Bundesrat, legislative powers, in terms of policy design (such as the drafting of legislation), are frequently in the hands of the federal government. At the same time, most public tasks are executed at the Länder or local level (Mayntz 1999, 101;Kropp 2010, 120). Hence, policy design is traditionally located primarily in federal bureaucracy, whereas authorities at the state and local level are mainly responsible for policy implementation. However, these classic notions of German executive federalism are partly set aside in the Online Access Act (OZG) including all three administrative levels as well as external actors in the policy design process, while simultaneously focussing on policy implementation.
The Online Access Act obliges all federal levels to offer their administrative services digitally in a joint portal network by the end of 2022. Overall, 575 administrative services are affected by the OZG, most of which are carried out by the Länder and municipalities (469). These services are subdivided in the Digitalisierungsprogramm Föderal (federal digitalisation programme) into various life and business situations for citizens and the economy and then bundled into 14 subject areas (Federal Ministry of the Interior, Building and Community 2018a, 4). To adopt a user-centric approach, these subdivisions do not reflect formal competencies of the public administration, but have instead been drawn from the user's perspective (Federal Ministry of the Interior, Building and Community 2019, 15). As a truly novel means of executive collaboration, at least one federal ministry and one of the 16 Länder is jointly responsible for developing an implementation strategy for services of each subject area (Federal Ministry of the Interior, Building and Community 2019, 24-25). One interviewee stated that the allocation of the subject areas was regulated rather informally (personal interview no. 1.3).
In developing an implementation strategy for two selected, high-priority services, the subject areas are supported by so-called digitalisation labs. These labs are temporary governance arrangements, financed by the Federal Ministry of the Interior, Building and Community (BMI) and organised by consultancies. The digitalisation labs include experts in the field of digitalisation, technical and legal issues from across all federal levels, as well as users of the particular service (Federal Ministry of the Interior, Building and Community 2019, 26). The main aim of the labs is to contribute to a more userfriendly implementation of administrative services (Federal Ministry of the Interior, Building and Community 2018b, 7-8) and to integrate all federal levels in the design of these services (personal interviews no. 1.2 & 2.1). At the end of a digitalisation lab, a user-friendly target process in the form of a detailed process modelling and a click-dummy, as well as a plan for implementation and proposals for necessary legislative changes should be developed (Federal Ministry of the Interior, Building and Community 2019, 34). The implementation solution developed in the digitalisation labs will then be made available to the other Länder in order to harmonise the provision of services across all municipalities (Federal Ministry of the Interior, Building and Community 2019, 95).
However, the labs are equipped with less formal authority than traditional forms of bureaucratic decision-making. Decisions taken in the laboratories have no legally binding effect. Although the Länder and local authorities are to be encouraged to implement the services developed in the laboratories in a uniform manner, the actual implementation will later be the responsibility of the legally responsible administrative unit. Similarly, the labs can make proposals for necessary changes in legislation and formulate possible legislative proposals, but they are not directly involved in the legislative decision-making processes. The laboratories and the OZG implementation process might therefore be described as a Type II governance structure, embedded in the general Type I governance of the German federal state. Nevertheless, the results and proposals of the digitalisation labs have a certain pressure potential on the other Länder and legislative bodies. The Länder do not have the capacity to digitalise all administrative services of the OZG individually and are therefore partly dependent on the results from the laboratories (personal interviews no. 1.1, 1.2 & 1.3). Similarly, this would entail high transaction costs for the legislative actors to nullify the laboratories' results.
The Online Access Act in general and the labs in particular are characterised by functional differentiation and decentralisation and focus on problems rather than on formal administrative competences. Various actors are involved in the policy design process in order to find a common solution for a concrete policy problem -the digitalisation of one administrative service. These institutional settings promote decisionmaking behaviour in the labs that is oriented rather towards problem-solving over bargaining. Or, to put it in other words: […] we play the three musketeers. One for all, all for one. (personal interview no.

1.2)
Organisational Design Different strategies are used in the laboratories to promote policy design and problemsolving. These strategies result both from the organisational design of the laboratories and from the composition of the participants. Using modern methods such as Scrum and Design Thinking, the labs are to develop implementation plans and target processes (Federal Ministry of the Interior, Building and Community 2018a, 7). Aside from the general question of the right governance tools, agenda setting, in the form of a concrete problem definition, plays an important role concerning policy solving capacity (Thomann, Trein, and Maggetti 2019, 39-40). To facilitate problem-solving as the first step, the actual state of the service is determined in the digitalisation labs. The different actors should contribute with their expertise from diverse perspectives to identify the various facets of the problem. At the same time, the joint definition of the problem strengthens the laboratories' focus on developing a solution for a specific policy problem. Based on problems identified in the actual state, in the second step the lab, participants then jointly develop a target vision for the service (Federal Ministry of the Interior, Building and Community 2019, 34). In contrast to classical policy design and decision-making processes, the actors are explicitly encouraged to develop solutions regardless of any legal or financial constraints (personal interview no. 1.1). This should strengthen the focus on problem-solving and postpone potential conflicts. This strategy of conflict avoidance plays a particularly important role in the laboratories. Certain conflicting topics are postponed to the end of the labs or excluded completely. It is made clear to the participants that the focus in the labs should not be on legal issues, but on the digitalisation of services. Questions relating to resources and competencies should, depending on the design of each lab, be postponed or not addressed at all (personal interview no. 4.1).
We would try to keep resource topics out of the labs a little bit, because we do not want to burden the participants unnecessarily. They should really concentrate on outlining a cool, new, innovative, citizen-friendly way of making use of administrative services. And they should explicitly not deal with the question of how much this would cost. (personal interview no. 4.1) Furthermore, the political implications of the concrete implementation should not be discussed in the labs. The participants should mainly deal with how the service can be designed as user-friendly as possible. This postponement of conflict might facilitate the policy design process in the labs, but consultants organising the labs particularly acknowledge the potential weaknesses of this method (personal interviews no. 4.1 & 4.2). This method deliberately keeps participants from knowing about certain hurdles to foster problem-solving. They must therefore ignore the fact that the developed solutions may not be legally feasible or that some relevant databases may be missing. However, these obstacles remain outside of the labs. For example, necessary legislative changes may have the potential to delay the implementation process for years. One consultant interviewed worries that participants are not sufficiently aware of this danger and therefore underestimate existing hurdles (personal interview no. 4.1). Additionally, some actors expect that problems may arise at the end of the labs as a result of the postponement of conflicts, therefore endangering the labs' results. If issues such as legal changes, resources, distribution of tasks or safety aspects are ultimately addressed, there is a risk that previously made decisions will no longer be supported by all participants. Furthermore, some interviewees expect problems when initial results are brought to the ministries or the public (personal interviews no. 1.1 & 4.1). This might be especially so since the results of such a salient topic as 'Immigration and Emigration' could be strongly politicised (personal interview no. 4.1). Since the laboratories do not decide on the distribution of resources, the classic mode of 'distributive bargaining' can be initially avoided. However, long term conflicts may arise if changes in legislation or distribution of funds are implemented.
Another strategy of the laboratories' ambition to promote problem-solving is the reduction of decision alternatives. After the participants have had the opportunity to develop solutions to the policy problem they had identified, free of legal regulations or financial constraints, the consultants then channel the ideas from the labs and develop a click dummy; a prototype for a frontend solution. Subsequently, the discussions can take place based on the click dummies developed by the consultants. The participants can then click their way through the prototypes and give their feedback (personal interviews no. 1.1, 4.1 & 4.2). In the interviews, the consultants made it clear that they would not be able to incorporate all ideas from the labs into the click dummy in the same way.
There will also be a lot that will be omitted, […] where ideas, things, requirements, tasks will be recorded, but first de-prioritised and put on the back burner. This should at least give people the feeling that their input is not lost. (personal interview no. 4.1) However, the consultants try to give the actors the impression that their input was relevant, even if not every idea is considered. This may affect the decision-making processes, as all participants feel that they have influenced the development of the prototype and therefore feel responsible for it. Also, by developing the click dummy, the consultants initially relieve participants of parts of the decision. As they do not allow certain ideas to flow into the prototypes, they may no longer be discussed in the labs.

Actors' Constellations
In addition to the institutional settings and organisational features, the constellation of actors in the labs has an influence upon the policy design process. In principle, the labs are composed of experts in the field of digitalisation, technical and legal issues from across all federal levels, as well as external actors such as users of the particular service, or other non-state actors (e.g. IT-providers, federal umbrella organisations). The concrete composition of each lab depends on the subject area and the service in question. Although a variety of different actors are represented in the digitalisation labs, the laboratories are limited to a small number of about 10 participants who are selected by the consultants (personal interviews no. 4.1 & 4.2). On the one hand, the diversity of the participants should strengthen the focus of the labs on the exchange of expertise and ideas, promote innovative problem-solving and facilitate learning processes. However, a reduction in the number of participants should facilitate the laboratories' problem-solving orientation by focusing on a small community of experts.
The different groups of actors take on different roles in the policy design process. While the national and federal actors are mainly seen as lab coordinators in the subject area 'Immigration and Emigration', the local level participants, users and external actors are mainly responsible for providing input. They contribute expertise on the implementation, technical realisation and use of the service and can provide insights into the concrete digitalisation processes. Simultaneously, the labs provide access to the policy design process for actors who were not previously involved or only entrusted with the later implementation. The labs thus open up a new arena of cooperation across federal borders, with the potential to initiate learning processes between actors and organisations. A special role in the laboratories is played by the consultants, who are responsible for scheduling, coordinating the actors and activities of the labs, as well as ensuring the progress of the project. They also develop prototypes for front-end solutions, so-called click-dummies from the results of the labs and support process modelling (Federal Ministry of the Interior, Building and Community 2019, 27). In doing so, the consultants have a great deal of freedom, since the BMI has given them very few fixed requirements for the design of the labs. The actual goals of the labs are fluid, the end product changes constantly and many terms used concerning the labs are not clearly defined. This gives the consultants a degree of influence over the labs' course and result (personal interview no. 4.1). Consultants also make important decisions about the composition and content of the labs. They can select the participants and set the agenda, therefore enabling them influence over the policy design process in the labs. In addition, the consultants prepare important information for the labs' participants, conduct user interviews and report their results to the lab. In doing so, the consultants can pre-select the topics to be discussed in the labs (personal interview no. 4.1). Some interviewees pointed out that the information provided to the participants has a big impact on the decision-making processes in the labs.
For example, the consultants send the slides and the concept […] and then we just look at it and basically agree to it.
[…] I also have the impression that we are not contributing much of our own ideas to it. (personal interview no. 1.1) In my opinion, of course, they somehow built a click dummy that they would have also built more or less even if the lab had not taken place. But at least the feeling that all input is important and evenly integrated into the end result, that is what they conveyed to the people very well. (personal interview no. 4.1) Concerning the constellations of actors, various strategies for promoting problemsolving in the laboratories can be observed. One of these strategies, the reduction of the number of actors involved, has already been briefly discussed. While many different groups of actors are supposed to be represented in the laboratories, the number of participants in the laboratories is kept low. This should facilitate the use of agile methods such as Scrum and design thinking, simplify collaboration between the actors and motivate participants to engage actively. The consultants want to give the participants the impression that they are of great importance for the process and that content can only be created in collaboration (personal interviews no. 4.1 & 4.2). This should lead to a group dynamic between actors that is oriented towards problem-solving, in which their own institutional origin plays less of an important role than the achievement of a common goal. Furthermore, the reduction of the number of participants leads to a specialisation of the participants. Since each organisation is represented by only a few participants, mainly public officials from the specialist areas, who have specialised either in the service under consideration or in the subject of digitalisation, are sent out to the labs (personal interviews no. 1.1, 1.2 & 4.1). Thus, because the group of participants is relatively homogeneous and mainly consists of experts in the field of digitalisation, the orientation towards problem-solving is strengthened, as the labs focus on professional exchange rather than on political debates.
Although the Länder are involved in the legislative process, they are usually left to their own devices when it comes to subsequent implementation. The digitalisation labs are intended to change this. They bring together representatives from across all federal levels to jointly develop an implementation solution for a common policy problem. This constellation of collaboration is not typical in the federal system and classical administrative structures and therefore represents a special, new form of cooperation (personal interview no. 2.1). The labs bring together people from a wide variety of institutions who have not previously exchanged ideas. Most of the interview partners consider it particularly important that the labs' actors work together on an equal footing, irrespective of their institutional backgrounds. This enables the free exchange of knowledge, experiences and interests across federal lines and could significantly improve exchange between federal levels (personal interviews no. 1. 1, 2.1, 2.3, 3.1, 3.2, 4.1 & 4.2). In addition to flat hierarchies, user orientation is of particular importance to collaboration in the labs. Different actors work together irrespective of formal responsibilities in order to design user-friendly administrative services. This common goal forces participants to abandon their rigid thinking about responsibilities. Looking back at Max Weber, 'responsibility' has always been a core term for the public administration. This jurisdictional responsibility no longer plays a significant role in the digitalisation labs. The labs focus exclusively on user orientation and problem-solving. According to some interview partners, this enables equal collaboration between the levels (personal interviews no. 2.2 & 2.3).
[…] the question of which department you come from, whether you are from the federal level, federal states or municipalities, is totally unimportant compared to the question of how we do it. (personal interview no. 2.3) However, while the labs may contribute to increased output legitimacy through potentially more adequate, problem-focused decisions and problem-solving strategies, concerns about their democratic accountability arise. As previously mentioned, the laboratories are mandated with comparatively few formal powers, as their decisions are not legally binding, nor are the laboratories directly involved in the legislative process. However, due to limited resources, both bureaucratic and legislative actors might be under a certain pressure to implement decisions and proposals from the laboratories or at least not to reject them completely. With regard to democratic accountability, it can be regarded as critical that actors who do not possess a democratically legitimised mandate at all or indirectly, gain influence on political decisionmaking processes. In this context, consulting agencies should be mentioned first and foremost. They have far-reaching decision-making powers in the laboratories as they can decide on their agenda and the composition of participants as well as exerting influence on the decision-making processes in the labs. Moreover, the participation of selected citizens in the policy process can be criticised. Even if similar formats, such as deliberative mini-publics, are said to have positive effects on democratic legitimacy (Goodin and Dryzek 2006;Fung 2007), concerns arise regarding the representativeness of this type of participation. In contrast to classic forms of citizen participation, the labs are not open to all citizens in principle, but the users are recruited by the consultants themselves (often from their private environment). This reflects a further problem of the digitalisation labs; low transparency and visibility. Even if certain results of the labs, such as the click-dummies, are made available to the public, there is almost no public information about the processes in the labs nor composition of participants. CONCLUSION The analysis of the digitalisation labs in the subject area 'Immigration and Emigration' has shown that the institutional framework, organisational design and strategies, as well as the composition of the actors influence the labs' policy design process and promotion of problem-solving.
The implementation of the Online Access Act and digitalisation labs is characterised by a strong functional differentiation, decentralisation of policy design processes and a clear focus on problem-solving. In contrast to classical patterns of executive federalism, actors from all three administrative levels, as well as external actors, are involved in the policy design process. The focus is not on formal competencies, but on user orientation and the collaborative search for solutions to a concrete policy problem. This focus on problem-solving is additionally bolstered by the organisational design and the composition of the actors in the labs. To promote problem-solving, the consultancies rely not only on a joint problem definition but also on a strategy of conflict avoidance. Possible conflicting issues, such as legal or financial concerns, are mostly excluded in the labs in order to prevent distributive bargaining. This may facilitate decision-making processes within the laboratories, however, in the long term, conflicts might only be postponed, which could lead to problems in the later implementation of the developed solutions.
The digitalisation labs are a novel MLG arrangement, departing in many ways from classical patterns of the German executive federalism. They provide a new arena for bureaucratic and external actors to work together to find solutions to concrete policy problems. Formal competences and organisational affiliations are deliberately neglected in this process in order to promote problem-solving. Flat hierarchies, a focus on user orientation, and agile methods are intended to help break through classic paradigms in order to find innovative solutions to pressing policy problems. Another novel feature of this arrangement is that it gives actors access to policy design and decision-making processes that are not traditionally involved, such as the local level or non-state actors. On the one hand, this can promote the exchange of expertise and knowledge between the actors and initiate learning processes, while on the other, it raises problems regarding democratic accountability. Limiting the number of participants and focusing on user orientation are further strategies to promote problem-solving in the labs. The low number of participants due to labs typically consisting of relatively homogeneous groups of experts from the field of digitalisation, and therefore the collaboration in the labs is more about the exchange of professional expertise than about ideological or political discussions. The different actors in the labs are equally involved in finding a solution to a policy problem, regardless of their organisational affiliation. Formal competencies and objectives of the respective parent organisation take a back seat in the course of problem-solving. This uncoupling from the parent organisations' objectives may be further promoted by the fact that the labs are formally equipped with comparatively few competencies, since decisions have no legally binding effect.
This paper provides an initial scientific contribution about this novel form of collaboration in the German multilevel system and an in-depth insight into the institutional settings, organisational design and the actors' constellations of the digitalisation labs. It identifies various strategies to promote problem-solving in the policy design process and contributes to existing multilevel research literature by analysing this special form of collaboration, combining the analytical concepts of multilevel governance and problem-solving with empirical evidence from the German digital transformation. Since the empirical analysis is based on a case study of two selected cases, some findings might only be generalised to a limited extent. However, this article is a first attempt to shed light on a new governance arrangement, which has seen barely any scientific investigation. It holds the door open for future research.
Based on the results of this study, there are many new insights for a future research agenda, from both a theoretical and empirical perspective. While the paper uses a procedural definition of the analytical concept of problem-solving, thus focusing on the analysis of policy design processes, the study could be extended by an outcome-centred approach to problem-solving. In other words, it could be examined as to what extent the laboratories and the strategies used have actually contributed to innovative problem-solving as an outcome. Additionally, the question of the legitimacy and democratic accountability of such governance formats could be examined more closely. Furthermore, the empirical basis of the paper could be extended by analysing and comparing digitalisation labs from other subject areas. In doing so, one might be able to see how other factors, such as the organisational field or the salience of the subject area, influence the policy design process as well as the outcomes of the labs. Also, the role of external actors in the labs could be further examined. While in the subject area 'Immigration and Emigration' mainly bureaucratic actors were represented in the labs, other subject areas (e.g. building and housing) show a much greater involvement of non-state actors, such as IT-providers or federal umbrella organisations. A comparative analysis of the different composition of actors in the laboratories of various subject areas could provide more precise information as to the extent to which the concrete composition of the laboratories influences their processes and results. As the digitalisation labs have hardly been researched so far, the findings of this paper open the way to a broad range of subsequent research questions.