How multilevel governance interactivity can contribute to the robust governance of crises. Insights from the Covid-19 and the asylum crises in Europe

Abstract

This article investigates how interactivity within multilevel governance (MLG) contributes to the robust governance of crises, understood as the capacity to maintain core public functions while flexibly adapting and innovating under turbulence. Building on a conceptual framework that distinguishes vertical/intergovernmental and horizontal/state-society dimensions of MLG, the study identifies three prevailing modes of interactivity—coordination, cooperation, and collaboration. Through eight qualitative case studies of robust crisis responses during the COVID-19 pandemic and the 2015 asylum crisis across seven European countries, the article shows how different modes of MLG interactivity shape adaptive and innovative robust crisis governance. Findings highlight that coordination-driven MLG interactions predominated during COVID-19, supporting pragmatic adaptation, whereas the asylum crisis fostered multi-stakeholder collaboration enabling locally tailored innovation. Across all cases, informal, trust-based interactions proved essential to navigating uncertainty and enhancing robustness. By unpacking the modes through which multi-scalar and multi-stakeholder interactions unfold, the article contributes to a more grounded understanding of how MLG policymaking enables robust crisis governance.

Introduction

The robust governance of crises and of the conditions of turbulence deriving from them has become a matter of increasing scholarly attention, reflecting the exponential rise of emergency situations and the growing complexity of these crises. However, despite the valuable insights produced by various studies, the question of how to make crisis governance more robust, that is, capable of flexibly adapting and proactively innovating the modus operandi of government when confronted with upsetting challenges, remains unresolved (Ansell et al., 2023; Ansell and Trondal, 2018).

For a long time, the debate has predominantly focused on vertical interactions in emergency response, i.e., inter-governmental relations, and more specifically on the role of central governments in crisis governance (Schneider, 2011). Recently, however, scholars have emphasised the increasing importance of the horizontal dimension of governance interactions in emergency management—namely, the participation of non-public stakeholders (Parker et al., 2020; Ravazzi et al., 2025; Sylves, 2020). Our research contributes to this debate by focusing on multilevel governance (MLG henceforth), understood as an interactive mode of policymaking involving both public and non-public actors in multi-scalar and mixed policy processes. We ask: How do different modes of interactivity within MLG policymaking contribute to the robust governance of crises?

Torfing et al. (2025) argue that problem focused, multi-scalar interaction taking place in MLG policymaking “may help to circulate information and knowledge, reformulate problem understandings and common objectives, flexibly adjust public policy and regulation, and support local innovation based on joint learning and authorization of solutions tailored to local conditions” (p. 6). However, we still lack a clear understanding of how interactivity in MLG policymaking concretely works, and more specifically which types of multilevel interactivity, that is of relations taking place at the same time on the vertical/intergovernmental dimension and on the horizontal/state-society dimension of MLG policymaking, can favour the emergence of robust governance responses to crisis and turbulence.

To address this gap, this article elaborates a conceptual space of modes of interactivity within MLG policymaking based on the intersection between the vertical/intergovernmental dimension of MLG, regarding whether MLG interactions are initiated from above, by the national government or regional authorities with competence on a specific issue, or from below by local distributed actors, and the horizontal/state-society dimension, regarding the type of actors involved, whether primarily public or non-public actors. Three prevailing modes of interactivity can be identified, i.e. coordination, cooperation and collaboration. To understand how different modes of interactivity within MLG policymaking contribute to the robust governance of crises, we present the results of qualitative case-study research of MLG interactions in eight instances of robust governance responses to the Covid-19 crisis and to the 2015 asylum crisis across seven EU countries.

While looking at our eight cases of robust crisis governance, this paper aims to assess which mode of MLG interactivity emerged and how it contributed to robustness in the context of crises that present distinct genesis and links with turbulence. While Covid-19 can be characterised as a situation of crisis-induced turbulence (Ansell et al., 2017), where a distinct event – the outbreak of pandemic infections – led to a wave of unpredictable effects, the refugee crisis constituted an instance of a creeping-crisis (Boin et al., 2020), the result of distinct events (the wars in Libya and Syria and the Arab Springs in Nord Africa to name the main ones) that generated mass arrivals of asylum seekers. Maladaptive policy responses by the EU and the more exposed member states, especially Italy and Greece, made in the summer of 2015 the situation untenable (see e.g., Geddes, 2022; Schilde and Goodman, 2021).

Research findings reveal that, in the context of the Covid-19 crisis MLG interactivity took primarily the form of coordination: interactions were started from above and regarded primarily public authorities and a limited number of stakeholders, interacting often informally and under the pressure to adapt national regulations and strategies to local conditions. In the case of the refugee crisis, instead, MLG was characterised by a greater involvement of civil society actors, with interactions often starting from below in order to shape locally tailored and innovative reception policies. In both cases, MLG interactivity was crucial in supporting pragmatic adaptation and innovation under crisis-induced uncertainty. The findings also underscore the important role of informality in facilitating the development of trust-based interactions not only between national/regional and local authorities but, above all, with civil society in the local communities confronting crisis and turbulence. While being exploratory in character, this study contributes to debates on robust crisis governance by proposing an empirical investigation of interactivity within multi-scalar and multi-stakeholder policymaking arrangements during two different and highly salient crises of our times, broadening our understanding of how interactivity in MLG policymaking can contribute to the robust governance of crises.

The article proceeds as follows. The next section reviews the relevant literature on robust crisis governance and MLG policymaking, introduces our conceptual space of MLG interactivity, and outlines expectations on the contribution of modes of interactivity to robust governance. We then present our research design, methodological approach, and the selected cases. This is followed by an analysis of eight cases of robust responses to the COVID-19 and asylum crises. Finally, we discuss the key findings, highlight the study’s contribution to ongoing debates on MLG policymaking and robust governance, and propose directions for future research.

Interactivity in multilevel governance as a driver of robust crisis governance

In the era of poly-crises, characterised by multiple and often interconnected disruptive events that interact in ways that amplify one another (Lawrence et al., 2024), the turbulence that derives travels rapidly across governance levels and institutional boundaries, making multilevel governance (MLG) a key condition for the robust governance of crises, defined as the capacity to uphold core public functions, goals, and democratic values while flexibly adapting and proactively innovating the modus operandi of government (Ansell et al., 2023). Going beyond resilience, understood as the restoring of the status quo ante (Windle, 2011), robust governance aims to achieve a dynamic balance of stability and change, adaptation and innovation. Multi-scalar and multi-stakeholder interactions are crucial to this end, since they can contribute to favour the rapid circulation of information, enable smooth decision-making, and set the conditions for adequate implementation (Torfing et al., 2025). However, how interactions within MLG policymaking concretely unfold—particularly under the uncertainty, complexity, and time pressure that characterize crises and the associated turbulence—remains insufficiently understood.

MLG and robust crisis governance

Debates on MLG have long been focused on definitional disputes, with scholars emphasising either institutional structures or policymaking processes (Papadopoulos et al., 2024; Tortola, 2017). These different conceptualizations also underpin the scarce existing research on MLG in crisis management, leading to contrasting findings. Studies that define MLG as an institutional multilevel structure – essentially synonymous with federalism and other decentralized systems – highlight a slower response of MLG systems to the challenges of COVID-19 (Carroll et al., 2025). In contrast, research that conceptualises MLG as a specific mode of policymaking, underlines the importance of multi-scalar and multi-stakeholder interactivity to crisis management (Newig and Fritsch, 2009; Klinke, 2017. More specifically, MLG has as a specific mode of policymaking implies that 1) different levels of government are simultaneously involved, 2) non-governmental actors at different levels are also involved and 3) relationships take the form of non-hierarchical networks between autonomous and interdependent actors (Piattoni, 2010).

Taking this perspective, Torfing et al. (2025) looks at MLG policymaking as a factor contributing to the robust governance of crises, favouring both flexible adaptation and innovation. The continuous engagement between higher-level government actors and lower-level authorities on the vertical/intergovernmental dimension of MLG can facilitate the flexible adaptation of laws, regulations, and recommendations to continuously changing conditions, while also allowing the garnering of support for exceptional local actions (Torfing et al., 2025). At the same time, interactions on the horizontal/state-society dimension are considered crucial for policy innovation, since they enable the inclusion of locally embedded knowledge and networks in crisis response (see also: Newig and Fritsch, 2009; Klinke, 2017; Agranoff, 2018).

As is clear, the emergence of robust governance solutions in the face of crises depends significantly on the quality of MLG interactivity, that is, on the nature of relationships among distributed actors interconnected through complex and multi-scalar interdependencies. Homsy and colleagues (2019) identify five key ingredients to ensure a functioning MLG process: sanctioning and coordinating authority, provision of capacity, knowledge co-production, framing of co-benefits, and engagement of civil society. This resonates with crisis governance literature emphasising that not all governance responses to crisis happen via the official chain of command, and that coordination should leave enough room for the incorporation of contextual knowledge and the empowerment of local actors (Sylves, 2020). Several studies demonstrate that unplanned strategies, non-codified behaviours, self-organization, and informal or underspecified structures can become essential in addressing crises (Ansell et al., 2010; Kendra and Wachtendorf, 2003; Ravazzi, 2024; Schomaker et al., 2021). More importantly, informality is a typical mode of interaction employed by central emergency management teams when seeking information from lower levels of government or coordinating operations (Waugh, 2000; Arendt and Alesh, 2015; Sylves, 2020).

However, this virtuous combination of formal and informal interactivity within the MLG policymaking of crises cannot be taken for granted. Typically, crisis governance implies trade-offs between speed and accountability, centralisation and decentralisation, coordination and improvisation (e.g., Boin et al., 2020). With respect to MLG, critics note that the multiplicity of actors and interests involved in multi-scalar and multi-stakeholder interactivity can lead to high levels of fragmentation in decision-making and implementation (Allen et al., 2023), while informality can result in opaque procedures, raising concerns in terms of democratic legitimacy and accountability (Papadopoulos, 2010; Sylves, 2020). A new conceptual framework is therefore needed to understand how interactivity concretely unfolds within MLG policymaking, and, most importantly, which types of MLG interactivity can contribute to the robust governance of different instances of crisis.

Conceptualising interactivity within MLG arrangements

To build this framework, we elaborate on the so-called “3-C” underlying inter-organizational arrangements (Keast et al., 2007), i.e. coordination, cooperation and collaboration, to conceptualise them as more general modes of interaction between public authorities and non-public actors in MLG policymaking arrangements. More specifically, we look at how these modes of interaction concretely unfold within MLG policymaking. To this end, we identify two key analytical dimensions of MLG policymaking: the direction of steering and the type of actors involved. We argue that the kind of MLG interactivity taking shape at the intersection of these two dimensions can be more or less conducive to the emergence of robust governance responses to crises.

The first dimension regards the direction of steering, i.e., who sets the goals and drives the emergence of multi-scalar interactions during the unfolding of the crisis. MLG interaction may be started from above, when international/supranational institutions and/or national or regional governments with competence on the specific crisis issue take the lead and steer top-down decision-making in response to the situation. Alternatively, MLG interactions can be initiated from below, through the actions put in place by local governments and stakeholders to face the specific challenges experienced by local communities, (Agranoff, 2018). The second dimension refers to horizontal/state-society relations, and more specifically to the types of actors involved in MLG policymaking. We distinguish between “primarily intergovernmental” MLG policymaking, involving mostly governmental actors and a limited number of stakeholders, and “multi-stakeholder” arrangements, including a broader range of societal actors. The prevalence of either type shapes the level of multivocality in crisis responses (Torfing et al., 2025: 6), that is, the capacity to incorporate diverse perspectives and audiences, thereby favouring innovation in crisis governance.

As shown in Figure 1, the two dimensions define a conceptual space where the “3-C” modes of interactivity can be located. These modes should not be understood as rigid types, since boundaries are typically blurred and empirical cases often can fall at the intersection between different modes of interactivity. Rather, the conceptual space offers a heuristic to deepen our understanding of how MLG interactivity can take shape, and enables to formulate expectations on how different modes may contribute to the robust governance of crises in terms of flexible adaptation and innovation.

Figure 1.

Modes of MLG interactivity. Conceptual space.

When MLG policymaking arrangements are started at the top by a governmental institution/agency (e.g., international/supra-national organisation, national/regional government, etc.) and participants are primarily representatives of other public institutions and a few stakeholders (non-governmental organizations and/or private companies), multi-scalar interactivity is typically characterised by coordination, in the sense that higher level authorities decide to pursue a certain strategy in response to the crisis and, to this end, start relations with local administrations and key stakeholders to align their strategies, tasks, and resources with the established goals. Coordination is a very common mode of multi-scalar and multi-actor interactivity in crisis management, especially when time pressure is high, such as in situations of threats to national security or natural disasters. Leadership is typically concentrated at the top—at the national or federal level (e.g., the prime minister or president with her staff)—and interactions are limited to key civil society organizations such as the Red Cross and/or a few large corporations (Waugh, 2000; Waugh and Streib, 2006; Somers and Svara, 2009; Schneider, 2011; Kapucu and Boin, 2016; O'Donovan, 2019). While coordination usually operates as a formal and hierarchical arrangement (Peters, 2021), under crisis conditions it also frequently unfolds through informal interactions, facilitating more rapid responses and allowing actors to flexibly adapt centralised measures to continuously changing and unpredictable challenges (Ali et al., 2025).

When MLG policymaking originates from below—meaning that it is started by local authorities and/or other grassroots actors seeking to find solutions to the crisis situation in their communities, interactivity tends to take primarily the form of cooperation, intended as the exchange of information and the search for recognition and support to carry out the actions that are deemed as more appropriate to solve specific local problems (Agranoff & McGuire, 2003). Cooperation in MLG arrangements is generally informal and based on goodwill, reflecting the voluntary joint actions that otherwise autonomous organisations undertake in order to face a challenging situation (Keast et al., 2007). In terms of contribution to robust crisis responses, it may be expected that activation from below and the prevalence of informal relations will favour the adaptation of national/regional-level responses to the specific conditions of local communities, and also enable innovative thinking when diverse local stakeholders are involved.

Beyond cooperation, interactivity within MLG policymaking can take the form of collaboration when public institutions at the top engage with other public administrations/agencies and non-public stakeholders at different scales in a collective decision-making process (Ansell et al., 2023). In other words, collaboration within MLG policymaking requires that distributed actors across different levels of governance converge in a process of policy co-creation, whereby policy goals are not set in advance from above or from below, but are decided jointly. In multilevel settings and in face of crises, co-creation will typically take place through the setting up of a specific venue or another institutionalized partnership steered by a public governmental authority. Collaborative interactions within these formal structures are likely to be non-hierarchical and informal, based on trust, shared norms, and dialogue. With respect to robust governance, the inclusion of diverse stakeholders across levels of governance is crucial not only for ensuring flexible adaptation but also for fostering innovation, as observed in research on COVID-19 emergency management processes (Ravazzi et al., 2025).

Hence, interactivity within MLG policymaking characterised by multi-scalar and multi-stakeholder arrangements can take different forms and can contribute to robust crisis governance by favouring adaptation and innovation in varying degrees. These different forms of interactivity not only, following Keast et al. (2007), imply a greater investment in relationships, with collaboration being particularly relations-intensive and coordination and cooperation less so, but more importantly in a context of crisis, they imply different degrees of involvement of governmental actors and stakeholders in identifying possible responses, and therefore in designing the goals and very rationale of crisis governance. In other words, coordination typically presents pre-set goals from above, with some limited room for innovative ideas coming from other actors; cooperation is actively sought from below with the goal of adapting policies to local conditions; collaboration is a mode of joint-decision, where goals are not set in accordance to an already established policy, but are co-created together, allowing a greater margin for innovative solutions that reflect the perspective of the various – public and non-public – actors that at different levels are facing the crisis situation.

In the following section, we apply our conceptual framework to eight cases of MLG policymaking that led to the robust governance of two highly salient crises of our times—i.e., the COVID-19 and the refugee crises. These crises were characterised by a specific genesis and links with turbulence. Whereas in COVID-19 turbulence followed a distinct event—namely the outbreak of pandemic infections—generating a situation of crisis-induced turbulence (Ansell et al., 2017), the refugee crisis constituted an instance of a creeping crisis (Boin et al., 2020), since it was the result of several events (the wars in Libya and Syria and the Arab Springs in North Africa, to mention just the main ones) and the maladaptive policy responses that followed (Geddes, 2022; Schilde and Goodman, 2021). Through a qualitative and inductive approach, we assess whether interactivity in MLG policymaking took different modes in the two crises and how different modes contributed to robust governance overall.

Case selection and methods

To answer our key research question regarding which mode of MLG interactivity can favour the emergence of robust governance in times of crisis-induced turbulence, we develop a qualitative multiple case study, based on the in-depth analysis of several cases with the objective of drawing inductive insights across different settings or contexts while treating each case as a self-contained unit.

More specifically, we focus on eight local cases of robust governance response to two main crises that affected the EU in the last decade, namely the Covid-19 pandemic (five cases) and the asylum crisis in 2015 (three cases). Building upon the rich data collection carried out within the EU Horizon Europe project Robust – Crisis Governance in Turbulent Times (GA 101061272) on responses to these crises in eight European countries (Belgium, the Czech Republic, Denmark, Estonia, Hungary, Italy, the Netherlands and Spain), case selection has proceeded in two steps. In the first step, we collaborated with the project partners responsible for research in each country to identify cases of robust governance responses to the two crises. To assess their robustness, we focused on levels of adaptation and/or innovation of governance responses. 16 robust cases were selected at the initial stage, but four were later excluded after fieldwork, as their robustness was not as evident as initially assumed. In the second step, we identified, among the 12 robust responses, those characterized by the presence of MLG policymaking – i.e., those exhibiting the three key features of MLG in terms of: 1) simultaneous engagement of different levels of government (at least two levels of government); 2) involvement of non-governmental actors at different levels; 3) relationships taking the form of non-hierarchical networks (Piattoni, 2010). We ultimately identified eight cases of subnational MLG policymaking processes that were associated with robust crisis governance.

Our analysis cannot claim to make any causal inference on the links between MLG and robustness, since this would require a rigorous selection of non-robust governance cases as well. This study clearly takes an explorative approach, with the goal of observing how different modes of MLG interactivity contributed to the robust governance of crises and the turbulence they generate in terms of policy adaptation and/or innovation. By examining multi-scalar governance responses to two distinct crises (the Covid-19 and asylum crises) our analysis also offers insights into how interactivity within MLG policymaking can take different forms in response to specific challenges.

The data collection process was coordinated by the authors of this article in collaboration with the Horizon Europe Robust Research Consortium and involved collecting both interviews and documents (media articles, policy documents, reports etc.). Interviews were conducted in August-October 2023 by field researchers in the local language, following a joint case study protocol and interview guidelines, and were aimed to reach all of the key stakeholders involved in policymaking process, i.e.: representatives of national, regional and local public sector decision makers; public health agency manager; politicians/administrators; civil society organization staff; non-public organizations; experts and task force members; journalists. Overall, 55 interviews were conducted, with an average of 7 for each case.

Interviewees were initially asked questions designed to gather information on the crisis and the responses implemented (pre-crisis preparedness, turbulence during the crisis, crisis responses). They were then asked specific questions on MLG relations and interactivity within these relations in crisis governance. Key relevant parts of the interviews were translated in English and data were analysed applying qualitative content analysis based on a joint case report form prepared by the authors of this article. Due to the complexity of working with different languages, the first stage of the data analysis was decentralized to local research units, who analysed data using software for qualitative data analysis and produced a detailed report about each of the case studies. This work was conducted under the coordination of the authors of this article, who maintained regular communication with the local teams and sought clarification on any unclear points.

Findings

Below we briefly present the evidence collected on interactivity in instances of MLG policymaking arrangements developed in response to the Covid-19 and the asylum crises, and on how these interactions contributed to the robust governance of these challenging situations.

Covid-19 crisis

Case 1 (Denmark): Vaccination in the Copenhagen Region and the case of Ishøj

The vaccination campaign in Denmark is representative of a multilevel governance arrangement started by the National Health authority and coordinated by regional health authorities, where intergovernmental relations were prevalent. In the case of the Copenhagen Regions, interactions took primarily a coordination role, in the sense that policy regulations, resources and information flowed from national to regional authorities and then to municipalities, with little feedback upward. However, in Ishøj, a municipality with one of Denmark’s highest COVID-19 incidence rates and one of the largest proportion of immigrants (around 50%) many of whom sceptical about vaccines, a remarkable case of cooperative interactions emerged, with local authorities involved in a direct dialogue with the Copenhagen Region on how to organize vaccination locally and mobilise grassroots networks. More specifically, multiple municipal departments (health, eldercare, schools, housing, and culture) coordinated laterally, while also collaborating with community actors such as imams, housing associations, and civil society groups. The municipality’s Cultural Centre was identified as a possible additional vaccination centre that could complement regional ones, and a local agreement was signed with the Region to this end. Interactions that developed on both the horizonal and vertical dimensions of MLG policymaking were reportedly very informal. The response can be considered robust as it allowed to curb disease spread and mortality, managing to uphold core public services via fundamental adaptations to the local unfavourable conditions.

Case 5 (Spain): Vaccination in Aragòn

MLG interactions underpinning the vaccination campaign in the Spanish region of Aragón were started from above, by the Autonomous Community in charge of implementing the national vaccination strategy, and involved primarily, even though not exclusively, other public actors. At a regional level, the vaccination process and prioritization were managed by a multidisciplinary crisis committee established by the regional government and composed of 12-14 members including representatives of the medical and nursing departments, and non-public stakeholders like technicians, veterinarians, pharmacists, logisticians. In the organization of the vaccination process, a major problem was that of reaching out to small localities and to migrants working in the agricultural sector. The regional civil servants interviewed stated that the opinions and suggestions of mayors in small municipalities and of neighbourhood associations were always taken into account. In rural areas with a very low population density, civil servants met with local authorities on a regular basis via WhatsApp groups. Furthermore, they engaged with imams and other religious leaders to persuade hesitant migrant agricultural workers. Hence, different types of actors were separately consulted by the regional government via formal channels and informal meetings and phone calls to coordinate at best the vaccination effort. Overall, this qualifies as an example of MLG arrangement that crucially contributed to the robust governance of the vaccination campaign in Aragon, since it allowed to overcome logistical constraints in dispersed rural areas and to reach vulnerable and hesitant migrant groups.

Case 3 (Belgium): ELZs and contact tracing

The case of contact tracing response in Flanders region reflects a MLG policymaking arrangement started from above that favoured the engagement, at a local level, of multiple stakeholders. It unfolded within a highly complex multilevel system, where the regions are primarily responsible for infectious disease control, but where the federal level played a major coordinating role. In fact, in April 2020 the Inter-federal Committee for Testing and Tracing was established in order to ensure a centralised coherent strategy and harmonised procedures. To implement this strategy, in September 2020 the Flemish government assigned the task of contact tracing to the Eerstelijnszones (ELZs), i.e., networks composed of primary healthcare providers, representatives from local authorities, insurance companies, user associations, and experts. Being responsible for a certain local area, each ELZ possessed extensive local knowledge and trust from residents – all essential advantages in the case of contact tracing. ELZs characterised as multi-stakeholder and highly institutionalised networks, since the regional Health Council was in charge of guiding the activities. However, interactions within the ELZs were reported by interviewees to have been collaborative and highly informal, thanks to pre-existing trust relationships. Hence, the ELZs contact tracing initiative developed by the Flanders regional government emerges as a remarkable illustration of robust governance pursuing the flexible adaptation of the federal strategy to local specific conditions, enabling the reaching of vulnerable populations and the enhancing of social trust.

Case 4 (Czechia): Crisis committees and the Brno’s regional vaccination centre

Regional Crisis Committees in Czechia are illustrative of an MLG policymaking arrangement formally established from above by the national government to implement central guidelines and coordinate crisis management in their territories, but that de facto favoured the activation from below of primarily public actors in a context characterised by flawed and inconsistent national regulations. In fact, regional crisis committees became important MLG venues where representatives of the local and regional executives and nongovernmental actors gathered and concretely shaped policy solutions. The interviews indicate that the composition of people participating in these platforms varied during the pandemic and often reflected the topic that was being discussed, including primarily representatives of public agencies like heads of hospitals, social services, territorial departments of police or fire brigade, and health insurance funds. Other non-public actors joined the meetings ad hoc, such as, NGOs, groups of volunteers, and professional associations of GPs on discussions regarding vaccinations. Interactions were mostly informal and collaborative, favouring robust governance in terms of flexible adaptation and innovation, as highlighted by the Brno’s regional vaccination centre, that became a model for the other regions. The centre was created within days by repurposing a temporary field hospital at the Brno Exhibition Centre thanks to the collaboration between the University Hospital, the municipality, Brno Trade Fairs, regional authorities, the fire brigades, the police and a large number of local groups of volunteers.

Case 5 (The Netherlands): Vaccination in the case of Urk

The case of the vaccination campaign in Urk stands for a case of MLG policymaking started from below, i.e, by the municipality, seeking coordination with higher level authorities in order to obtain recognition and support for the informal collaborative network that had been developing in the local community in order to increase vaccination rates. It therefore involved primarily public actors. In fact, in Urk the national large-scale vaccination strategy, based on the opening of a big vaccination centres located outside the village, was met with strong hostility mostly due to religious motivations, leading also to the staging of demonstrations. While local authorities were struggling, general practitioners (GPs) offered to vaccinate citizens within their practices, positioning themselves as trusted, local professionals. Initially, the initiative faced challenges due to a lack of communication between the municipality and the regional Public Health Service (PHS). The mayor used existing platforms, such as the ‘Safety region’ and direct interactions with the Minister of Justice and Security, to draw attention to the specific situation in Urk. Deliberations were mostly ad-hoc and took place through informal channels, such as phone calls between the local government, the Minister of the Interior, and the Ministry of Justice and Security. Over time, communication was established among the different actors. GPs were put in the condition to administer vaccines, while the PHS coordinated the logistic, to ensure that all clinics received the necessary number of vaccines each day and that they were stored in safe conditions. Multi-scalar cooperative interactions were therefore crucial in ensuring the robust governance of vaccination in Urk, allowing for the adaptation of the vaccination strategy to the unique characteristics and needs of the local community.

Asylum crisis

Case 6 (Italy): The Tuscan reception system #AccoglienzaToscana

To face the necessity of finding suitable accommodations for asylum seekers, a MLG venue was established in 2015 from above by the regional government of Tuscany, gathering together municipal governments (including the regional branch of the National Association of Municipalities – ANCI) and key non-governmental organizations (ARCI, Caritas, Oxfam; for more details see: Pettrachin, 2022). According to interviewees, interactions were facilitated by the fact that the actors involved had similar political ideologies, since the regional government and municipalities involved were all from centre-left parties and they relied on established patterns of cooperation with left-wing or progressive Catholic civil society organizations. This collaborative multilevel interactivity favoured the robust governance of the crisis in the Tuscan region through the setting up of an innovative reception system based on small structures dispersed throughout the whole regional territory. The direct engagement of NGOs in designing the system enabled the promotion of a grassroot work of dialogue with the local population in order to facilitate the acceptance of reception facilities.

Case 7 (The Netherlands): Plan Einstein

This is a case of a bottom-up and multistakeholder MLG arrangement in response to the national government request to expand municipal reception capacity during the 2015 refugee crisis. After forming a network with diverse partners in Utrecht, the city government developed Plan Einstein (formally: Utrecht Refugee Launch Pad), repurposing an office building in the historically diverse and disadvantaged Overvecht district into an asylum shelter hosting 400 asylum seekers alongside 38 young residents from Utrecht (see: Dekker et al., 2021). Notably, the educational and recreational services of Plan Einstein were accessible to asylum-seekers, as well as to members of the local community. The horizontal dimension of MLG is evident in the collaboration between local public authorities and NGOs, including national-wide organization such as the Dutch Council for Refugees, that contributed by sharing their knowledge and previous experiences regarding refugee reception. On the vertical dimension, the network applied for – and then obtained – EU funding to the Urban Innovation Action, and invited the national asylum agency COA to join the partnership. It preferred not to be official involved, even though at the local level, the COA manager and professionals of the reception centre fully participated in Plan Einstein, including attending consortium meetings (Dekker et al., 2021). Interactions within the MLG arrangement were initially highly informal, but later became formalised in the framework of an EU-funded project (Geuijen et al., 2020). The Plain Eistein project contributed to the robust governance of the refugee crisis in Utrecht by developing a locally tailored, innovative model of refugee reception that integrated co-housing, co-learning, and community engagement—far beyond the national government’s standard reception approach.

Case 8 (Belgium): The citizen platform (BELRefugees)

This is an instance of MLG policymaking developed from below by a civil society alliance that gradually established relationships of cooperation with public authorities at a local and regional level. In face of increasing arrivals of refugees in the city of Brussels, and of a lack of engagement by the federal and regional governments, local NGOs joined forces through social media and volunteers’ personal relationships. The Platform rapidly expanded into a large-scale civic actor capable of delivering essential services—food, shelter, medical care, and legal guidance. In this process, BELRefugees started to interface pragmatically with municipal and regional authorities. Brussels municipalities became involved first, cooperating with volunteers in order to provide sanitation, public spaces, and logistical support. Over time, the Brussels Capital Region, despite lacking clear competence, also offered limited material support, like allowing the use of public buildings or regional facilities for temporary shelter and providing financial resources to pay full time staff. BELRefugees illustrates how MLG policymaking originated in civic mobilization can contribute to robustness in crisis governance by facilitating decentralized problem-solving and the adjusting of operations to emerging needs, thereby stabilizing a situation in which the official (top-down federal) system failed.

Table 1 provides an overview of our findings on the eight cases of robust crisis governance, with specific attention to the two key dimensions characterizing interactivity within the MLG policymaking arrangements, i.e., the direction of multi-scalar policymaking and the type of actors involved, the mode of interactivity found as the prevailing one in each case, and the specific contribution to robust governance in terms of adaptation and/or innovation.

Table 1.

Interactivity in MLG policymaking in selected cases of robust crisis governance.

Case	Direction	Actors involved
Case 1 Vaccination (Ishøj, DK)	From above	Region, municipality + religious groups, NGOs	Coordination	Flexible adaptation
Case 2 Vaccination (Aragòn, S)	From above	Region, municipalities + NGOs, leaders	Coordination	Flexible adaptation
Case 3 Contact tracing (Flanders, B)	From above	Region, municipalities + multi-stakeholder	Coordination + collaboration	Flexible adaptation
Case 4 Crisis Committees (Czechia)	From above	Region, municipalities, other public + multi-stakeholder	Coordination + collaboration	Flexible adaptation + Innovation
Case 5 Vaccination (Urk, N)	From below	Municipality + GPs, NGOs + Region, National Gov	Cooperation	Adaptation
Case 6 Refugee reception (Tuscany, I)	From above	Region, municipalities + various NGOs	Collaboration	Innovation
Case 7 Refugee reception (Utrecht, N)	From below	Municipality + various NGOs + National officials, EU	Coordination	Innovation
Case 8 Refugee reception (Brussels, B)	From below	NGOs + Municipality, Region	Coordination	Innovation

Discussion

Figure 2 graphically represents the positioning of our eight cases of robust crisis governance within the conceptual space defined by the vertical and horizontal dimensions of MLG policymaking. The figure reveals, first and foremost, that robust responses were produced frequently at the intersection of different modes of MLG interactivity. In the following, we discuss the different modes inductively identified through our qualitative analysis, with the objective of drawing new insights on how MLG policymaking contributed to instances of robust crisis governance in terms of different degrees of adaptation and/or innovation.

Figure 2.

Modes of interactivity within MLG policymaking in the identified robust governance responses to crises.

The first mode of interactivity is that of coordination + cooperation, reflected by the MLG policymaking arrangements that were started by the Region of Copenhagen in case 1 and the Aragon Autonomous Community in case 2, which had primarily the goal of promoting the correct and efficient implementation of the national vaccination strategy. This required close and often informal interactions with local authorities, which had the opportunity to communicate difficulties and problems encountered in the local communities, and to engage non-public actors and leaders who could fruitfully cooperate in reaching particularly vulnerable or hard-to-reach populations like migrants or linguistic minorities. Overall, across these cases MLG interactivity allowed robust governance through the flexible adaptation of national strategies to local specific conditions that risked jeopardizing implementation.

Following clockwise, the second mode of MLG interactivity emerging from our analysis is coordination + collaboration. In Belgium (case 3) and the Czech Republic (case 4), contact tracing and vaccination strategies respectively were coordinated through ad hoc formal venues such as the ELZs in Flanders and the Crisis Committees in Czechia. In both venues regional authorities interacted from the outset with local authorities and key non-public stakeholders to shape concrete policy solutions. Collaboration prevailed in these venues, favouring robust governance in terms of flexible adaptation to the local conditions (case 3), and enabling the emergence of innovative solutions as pointed out by the establishment of the Brno vaccination centre (case 4).

The third mode of MLG interaction, prevalent collaboration, is exemplified by case 5 regarding the reception of refugees in Tuscany, where a specialised venue was convened by the regional government, #AccoglienzaToscana, bringing together municipalities and the regional branch of the national associations of municipalities on the one hand, and various NGOs already mobilised on migration issues on the other. Formal and informal relations among all the involved public and non-public actors were crucial in shaping a process of co-creation of an innovative, locally tailored refugee reception system, avoiding concentrations in big structures which had already emerged as highly problematic throughout Italy.

Last but not least, a fourth mode of MLG interactivity is that of prevalent cooperation. In this mode, either local authorities (cases 5 and 7) or NGOs (case 8) interacted horizontally to find solutions to situations of crisis, and then mobilised on the vertical dimension by seeking cooperation with higher-ranked authorities to obtain recognition and scale up locally robust practices. However, whereas in the case of vaccination in the Dutch locality of Urk (case 5), horizontal interactions were limited to municipal authorities, GPs, and volunteers seeking to adapt the national vaccination strategy to the local context, in cases 7 and 8 we find broader coalitions of non-governmental actors mobilised in shaping from below innovative responses to the challenge of refugee accommodation. In general, relationships at the local level were based on informality and trust, while to establish interactions with higher-ranked levels of government, both formal and informal channels were used by local actors. Informal interactivity appears particularly relevant to overcome political impasse in the highly politicized asylum issue.

With respect to the two types of crises considered in this study, some noticeable differences emerge when considering the direction of steering and the number and types of actors involved. During the COVID-19 crisis—which constitutes an instance of crisis-induced turbulence (Ansell et al., 2017)—MLG interactivity took the form of coordination: interactions were initiated from above and involved primarily public authorities and a limited number of key stakeholders, with the goal of adapting national regulations and strategies to local conditions. In the case of vaccination in the Dutch locality of Urk (case 4), where interactions were started by local actors, the underlying logic appears to be that of seeking cooperation with higher-ranked authorities because of the failure of the latter to engage in coordination. Overall, in the terms of Homsy et al. (2019), sanctioning and coordinating authority and provision of capacity appear as the key ingredients of MLG interactivity in crisis-induced turbulence, while knowledge co-production and framing of co-benefits seem to occur only in a second stance, and primarily in the coordination + collaboration variant. In cases 1, 2 and 5, where no collaborative venue was established, the engagement of civil society seems to have been primarily informal and focused on reaching out to vulnerable or hesitant groups, thereby increasing implementation capacity.

Regarding the refugee crisis—an instance of a creeping crisis (Boin et al., 2020)—we find that MLG was characterised by a greater involvement of civil society actors, with interactions often starting from below and leading, in the terms of Homsy et al. (2019), to knowledge co-production and framing of co-benefits, as highlighted by the shaping of locally tailored and innovative reception policies. The emergence of broader and more varied partnerships on asylum issues is not new, and reflects the significant role that civil society has traditionally played in mobilising in favour of migrants and refugees well before the 2015 crisis (Caponio and Ponzo, 2023). As for vertical relations, coordinating authority is certainly relevant but not indispensable for the start of MLG interactions, as shown in case 8; regarding provision of capacity, this was actively solicited from below to fill gaps in terms of funding and other material resources (notably in cases 7 and 8).

Research findings also underscore that interactivity within MLG policymaking contributes to the robust governance of crisis by promoting adaptation and innovation in different degrees. Flexible adaptation emerges as the main outcome of prevalent coordination and prevalent cooperation, whereas innovation appears more likely to occur when collaborative venues are established—i.e., in the cases of coordination + collaboration and prevalent collaboration, which include stakeholders in a process of decision-making and co-creation from the very outset of the crisis. At the same time, our study shows that across all three modes of MLG interactivity, including when a collaborative venue is formally established, interactive relations were always characterised by a high level of informality, suggesting that mutual trust, personal acquaintance, and improvisation are indispensable assets to MLG policymaking in times of crisis. These features ensure rapid communication, facilitate the exchange of experiences, and sustain mutual support across governance scales as actors confront the turbulence associated with crises.

Conclusion

This article set out to investigate how different modes of interactivity within MLG policymaking contribute to the robust governance of crises. By comparing eight cases across two distinct types of crises—the crisis-induced turbulence associated with COVID-19 and the creeping crisis of asylum reception—we developed an empirically grounded account of how interactivity unfolds across the vertical and horizontal dimensions of MLG, and how these patterns enable flexible adaptation and policy innovation. The study demonstrates that robust governance in turbulent times is not the product of institutional design alone but emerges through the dynamic interplay of public and non-public actors at different governance levels operating under uncertainty and time pressure.

Three core findings stand out. First, on the vertical dimension, multi-scalar interactions within MLG policymaking, whether initiated from above—as in many COVID-19 responses—or from below—as in several asylum-related cases— created positive conditions for the development of robust governance, providing opportunities for local knowledge to inform higher-level strategies, and for central guidance to be adapted to context-specific needs.

Second, the horizontal state–society dimension played a decisive role in determining whether MLG interactivity yielded robust governance in terms of flexible adaptation or more substantive innovation. In the COVID-19 cases, where interactions generally involved public actors and a limited number of key stakeholders, flexible adaptation prevailed. Formal and informal interactions between different public and non-public actors enabled adjustments to national strategies—such as fine-tuning vaccination delivery in rural Spain or engaging trusted intermediaries in Ishøj. By contrast, in asylum-related cases, multi-stakeholder involvement—including NGOs, volunteers, and civil society platforms—facilitated the emergence of experimental, socially embedded solutions such as Plan Einstein or the Tuscan dispersed reception model. These collaborative configurations reveal how broader actor diversity contributes to what the robustness literature identifies as multivocality: the capacity to incorporate heterogeneous perspectives into crisis response design (Torfing et al., 2025: p. 6).

Third, across all modes of interactivity, informality emerged as a fundamental driver of robustness. Informal communication channels, trust-based relationships, and improvisation enabled fast information flows, rapid problem-solving, and the bridging of institutional gaps. Even where formal venues were created, their functioning relied heavily on informal relations that facilitated collaboration and flexibility. This finding aligns with existing scholarship on informality in crisis governance (Sylves, 2020; Duda et al., 2020; Boin and Lodge, 2021; Ravazzi, 2024), while extending it by explicitly situating informality within multi-scalar and mixed-actor MLG policymaking arrangements.

Taken together, these insights deepen the scientific understanding of how robustness emerges in complex crisis settings. While our exploratory design does not support causal explanation, the consistency of patterns observed across diverse countries and across two distinct crisis types strengthens the robustness of our empirical findings, as similar configurations of MLG interactivity were associated with adaptive and innovative governance outcomes in markedly different institutional and political contexts. In particular, the study advances the literature in three main ways.

First, it conceptualises MLG interactivity as a multi-dimensional space, defined by vertical relations and actor composition, against which different modes—coordination, cooperation, collaboration, and their hybrids—can be identified. This conceptualisation offers a more nuanced understanding than binary distinctions between centralisation and decentralisation (Kuhlmann et al., 2024; Špaček, 2025) or formal and informal relations (Ali et al., 2025). Furthermore, our empirical analysis demonstrates that there is no unique path to robust governance, showing the crucial relevance of both multi-scalar and multi-stakeholder interactions in shaping different patterns of MLG interactivity for robust crisis governance. Second, it demonstrates empirically that different crisis types tend to elicit different modes of MLG interactivity, thereby enriching debates on the crisis–turbulence nexus (Ansell et al., 2023). Crisis-induced turbulence appears to privilege coordination-driven interactivity, whereas creeping crises create more room for bottom-up mobilisation and stakeholder collaboration. Third, it highlights the importance of informality and trust as cross-cutting elements underpinning the functionality and robustness of MLG processes, even within highly institutionalised settings. This helps reconcile tensions in the literature between the need for structured coordination and the acknowledged value of improvisation and self-organisation (Sylves 2020).

At the same time, this study raises several questions deserving further research. Future work could expand the comparative scope by including non-robust cases to identify necessary and/or sufficient conditions that underpin the emergence of an MLG-robust governance link in contexts of crisis. Moreover, systematic investigation of the temporal dynamics of MLG interactions—how modes of interactivity evolve as crises unfold—would advance our understanding of adaptive capacity and policy innovation over time. Quantitative or mixed-method approaches could also help assess the relative weight of vertical versus horizontal interactions in producing robustness across different institutional settings, and more specifically centralised versus federalist and decentralised systems. Finally, given the centrality of informality, future research should examine how trust-based relations can be fostered without compromising accountability and transparency (Sørensen and Ansell, 2021), especially in politically salient or polarised crisis contexts.

In sum, this study shows that MLG interactivity is not merely a structural backdrop to crisis governance—it is a critical policymaking arrangement through which robust governance can take shape. By unpacking the modes through which multi-scalar and multi-stakeholder interactions unfold, we contribute to a more grounded understanding of how public authorities and societal actors can jointly navigate the turbulence of contemporary crises.