EU free movement of people: fully recovered or suffering from long COVID?

ABSTRACT The free movement of people (FMP) was one of the first victims of the COVID-19 pandemic in Europe. Calls for better coordination often remained unheard in the initial phase of hectic border closures and mostly resulted in soft EU recommendations. Nevertheless, most restrictions on the FMP were lifted after summer 2020 and could be largely avoided in later waves of the pandemic. We argue that the resilience of the FMP, a fundamental principle of the EU, ultimately depends on policies – borders, welfare, and health – that are still national competence and partly implemented at the regional level. We explain the resilience of the FMP and the observed variation between different border regions by the strength of pre-existing coordinative transgovernmental networks. Our analysis is based on a comparative case study of different German border regions and draws on quantitative data concerning cross-border workers and qualitative interviews with policy-makers and stakeholders.


Introduction
Under the EU Treaties, people are entitled to move freely from one EU member State to another without suffering discrimination as regards employment, remuneration and other conditions of work. This free movement of people (FMP) principle was one of the first victims of the COVID-19 pandemic as member states installed rigid border restrictions to contain the virus. Wolff et al. (2020) showed how the prevalence of an internal security approach was shaped by previous experiences, notably the 2015 refugee crisis, and led to a re-nationalisation of border policies. Genschel and Jachtenfuchs (2021, p. 253) saw in this 'dramatic rebordering of the Single Market and the Schengen area' a sign that 'EU institutions lost regulatory control over free movement and have been unable so far to win it back'. Despite this gloomy assessment, others remarked on the rapid re-opening of the borders after summer 2020 and the increasing effort to coordinate responses to FMP disruptions (Thym & Bornemann, 2020).
Against this background, we ask how the re-opening of borders and the re-establishment of free movement was made possible. We focus our analysis on cross-border workers as they were the ones hit the hardest by the restrictions on free movement. Cross-border workers commute to work in another EU country on a daily or weekly basis and they are insured in the member state of employment. We argue that we should understand the disruptions to FMP not primarily as a conflict between member states or between EU institutions, but rather as a coordination problem in a complex field of multi-level governance that required a rapid and flexible response in times of high uncertainty. Indeed, the pandemic has shown how FMP, despite being a fundamental principle of the EU, ultimately depends on policiesborders, welfare, and healththat are still member state competence. Consequently, re-establishing and maintaining the FMP during the pandemic required a coordinated response across levels of governance, policies and member states. These different dimensions of multilevel governance help us to understand the way in which resilience in FMP was achieved and why it varied across border regions. We show how the resilience of FMP could be ensured by finding pragmatic and flexible solutions through transgovernmental networks in a multi-level setting and that the observed variation between different policies and border regions can be explained by the strength of pre-existing coordinative transgovernmental structures. For instance, while some regions found pragmatic solutions to coordinate test requirements and capacities, others struggled to ensure social rights (e.g., short-term [furlough] money) for those mobile workers who could not cross the border. These solutions came also from various sources: local authorities, regional and national governments as well as common EU solutions such as the Digital COVID Certificate.
The next section discusses the emerging state of the art on free movement during the COVID-19 crisis and presents our theoretical expectations. Subsequently, we introduce our case selection and method of data collection. We conduct a comparative case study of cross-border workers in different German border regions that is based on a triangulation of document analysis and expert interviews. Based on this, the empirical analysis in section four shows that the border closures and the reinstallation of free movement were not primarily attributable to EU-level decisions, but to transgovernmental coordination efforts at the level of member states and, significantly, regional authorities. The final section summarises the results and discusses their implications for EU multi-level governance in times of crises. forced many EU governments to repeatedly shut down large parts of public life. Strikingly, external mobility was sometimes even less restricted than internal mobility when people living in a country undergoing a lockdown were nevertheless allowed to travel abroad and move across EU borders with only limited constraints (e.g., passenger location forms, test and vaccination obligations being the most common). Apart from infection dynamics, we still need to understand better when and through which policies crossborder mobility was restricted (or facilitated) and how these policies affected the FMP in different border regions.

Freedom of movement from a multi-level perspective
To answer these questions, we draw on multi-level governance, since this approach allows us to theorise about how competences are distributed across levels of governance, policies and different regions and how actors participate in the problem-solving process (cf. Trein et al., 2019). Multi-level governance is generally defined as 'as a set of general-purpose or functional jurisdictions that enjoy some degree of autonomy within a common governance arrangement and whose actors claim to engage in an enduring interaction in pursuit of a common good' (Zürn et al., 2010, p. 4). Therefore, it comprises two different types of governance arrangements: Type I is composed of general-purpose jurisdictions with non-intersecting memberships that look like a Russian doll; Type II defines task-specific jurisdictions with intersecting memberships and flexible designs (Hooghe & Marks, 2010).
Type I governance can help us understand the opportunities and constraints that EU actors face in multi-level policy-making. To begin with, the distribution of competences across different jurisdictions provides an indication of who can become a focal point in the process of problem-solving. In our case, while free movement is a core goal of European integration, the measures to achieve itborders, welfare and healthare generally anchored at the national and subnational level. Indeed, they can all be considered 'core state powers' (cf. Genschel & Jachtenfuchs, 2016), since member states are reluctant to go beyond soft forms of coordination on the EU level, where the EU's competence remains mostly limited to supporting the actions of member states. EU institutions have better chances to become an orchestrator in areas where they can claim EU competences or where there are no other focal points acting as problem-solvers (cf. Abbott et al., 2015). For instance, in the area of health,  observed clear differences between a network on pharmacovigilance (i.e., linked to safety standards) and another dealing with cross-border health care (prone to distributive conflicts). Given the nature of these networks, the former was politically supported on the EU and national levels and emerged as a strong supranational network with close links to the Commission; in comparison, the latter received limited (national) support and developed into an information-sharing forum used mostly to mitigate negative externalities on the health care systems of member states. In contrast, Hollis (2020) offered the examples of German-Swiss emergency cooperation along the Rhine River and German-Dutch firefighter assistance to show that transborder crisis management was actually better prepared at the local than at the national and EU levels. Subnational actors were more concerned about finding efficient solutions and lowering transaction costs than about sovereignty concerns, which tended to plague cooperation at the EU level. Therefore, we expect the type of competences and the nature of the problem (coordination vs distribution) to determine the opportunities offered to specific actors situated at different levels of governance to get involved and be recognised as legitimate to deal with (a specific aspect of) the crisis.
Furthermore, Slaughter (2004b) showed that multi-level governance required a certain state of 'disaggregation' to be successful. The more 'disaggregated' states are, the more they provide the necessary space and flexibility for actors beyond the unitary state government to act autonomously. Indeed, the literature has shown that the expectation of a 'Europe of the regions' has turned rather into a 'Europe with some regions'; there are strong differences in the involvement of subnational actors depending on the extent of regionalisation in a country and the internal distribution of competences. Their engagement in EU politics is strongly correlated with their institutional capacities and interest in self-empowerment. This explains why the Eastern enlargement created a bigger gap between powerful regions in (quasi-)federal states and smaller regions operating in largely centralised political systems (Bauer & Börzel, 2010). Therefore, we expect the type of state governance (unitary vs federal) to explain variation in cross-border cooperation, since it provides actors at the borders with different degrees of autonomy and empowerment.
In the EU, Type II governance arrangements are embedded in a rather loose Type I structure. Indeed, Hooghe and Marks (2010, p. 25) highlighted that cross-border regions were perfect laboratories for Type II governance arrangements, since they tended to be functionally specific and prioritise problem-solving in collective action problems. They offered the example of the Upper Rhine Valley, bringing together the Swiss cantons of Basel-Land and Basel-Stadt, the French department Haut Rhin and the German region Baden as an example of a highly efficient transnational jurisdiction. These transnational jurisdictions are often managed by horizontal transgovernmental networks composed of public (governmental) actors working across borders and levels to solve common problems, especially when competences and resources are dispersed across various areas and levels of governance (Börzel & Heard-Lauréote, 2009;Slaughter, 2004a, p. 14).
Arguably, transgovernmental networks are particularly good at Type II governance for several reasons: (1) they create convergence and informed divergence (Slaughter, 2004a, p. 24) by offering the necessary flexibility and responsiveness to formulate and implement solutions in situations of uncertainty, high complexity and functional pressure (Börzel & Heard-Lauréote, 2009;Slaughter & Hale, 2010). (2) they improve compliance with Type I rules (Slaughter, 2004a, p. 24)including actors that are often directly or closely involved with implementation and who know the realities on the ground helps enhance the effectiveness of policy solutions, especially if the network deals with coordination problems rather than distributive conflicts (Börzel & Heard-Lauréote, 2009;Eberlein & Grande, 2005;Slaughter & Hale, 2010). (3) they 'increas[e] the scope, nature and quality of international cooperation' (Slaughter, 2004a, p. 24) over time. There is an expectation that close cooperation in networks will have a socialisation and learning effect, meaning that it will improve problem-solving and build trust among the members of the network (Börzel & Heard-Lauréote, 2009;Hobolth & Martinsen, 2013). Slaughter and Hale (2010, p. 367) also considered that 'building these networks in good times is creating valuable capital for addressing crises', since it allows actors to know who their contact partners are. This is particularly valuable for subnational entities, which gain a unique opportunity to socialise with their counterparts in other countries but also at different levels of governance.
In sum, we expect both horizontal and vertical transgovernmental networks to be particularly valuable in a crisis like COVID-19, which posed a symmetrical shock for all member states and raised problems of coordination rather than distributive conflicts (Quaglia & Verdun, 2023). Transgovernmental networks help to coordinate and lower transaction costs in times of uncertainty, notably by sharing information and best-practises as well as providing technical expertise that can serve as a basis for political decisions. Consequently, we expect that, in those policy areas where coordination is more advanced and established (vertical) networks can provide quick solutions to new problems, FMP will be more resilient. Finally, we expect that the type of political system together with the strength of pre-existing coordinative transgovernmental structures will explain the variation in how (and how successfully) FMP measures have been introduced in different border regions.

Case selection and method
As highlighted above, our empirical focus is on cross-border workers, since worker mobility is at the core of the FMP and they were particularly affected by COVID-19-related restrictions. German border regions were chosen for our comparative case study because, in terms of numbers, Germany is the favourite destination of cross-border workers (European Commission, 2021, pp. 70-71). Most cross-border workers come from Poland, the Czech Republic and France. As Figure 1 shows, while the numbers of cross-border workers from Poland and the Czech Republic have strongly increased during the last decade, the amount of workers from France remained largely stable (European Commission, 2021, p. 70). During the first wave of the pandemic, the number of cross-border workers from Poland and the Czech Republic dropped slightly (more than from France and independently from seasonal fluctuations), but, in the long-term the FMP proved to be resilient ( Figure 1, for more detailed figures, see Appendix II).
In addition to the country's attractiveness for cross-border workers, there are other reasons why German border regions are valuable cases for testing our theoretical expectations. First, as a federal country made up of 16 states, policy competences relevant for the FMP do not always lie in the hands of the federal level. This allows us to compare regional differences, e.g., varying entry and quarantine regulations imposed by the German states and different cooperation patterns between state and federal levels. Whereas the regulation and control of external borders is a federal competence, i.e., decisions are taken under the responsibility of the ministry of the interior and enforced by the federal police force, the distribution of competences in Germany is more complex regarding welfare and health. Social security is regulated at the federal level, but largely depends on the implementation by regional and local authorities such as the ten regional directorates and 156 employment agencies of the Federal Employment Agency. 1 Finally, the federal German infection protection law defines the available policy options, but their adoption and implementation are decided at the state level.
Second, located in the middle of Europe, Germany borders eight EU member states and Switzerland as a non-member. The different border regions experienced different challenges and trends in terms of worker mobility, diverse labour markets and infection rates. Finally, regional cooperation also varies due to Germany's historical past. While regional cooperation with its neighbours on the Northern, Western and Southern borders has a long tradition, regional cooperation with the Eastern neighbours is less well developed. Cooperation in three European Employment Services (EURES) networks linking national employment services and five border information points at the Dutch, Danish, French, Austrian and Belgian border often dates back to the end of the second world war. In contrast, there is only one EURES partnership with Polish and Czech regions and a loose partnership between employment authorities across the river Oder (Brandenburg-Poland border). 2 Data for the case study was assembled by a triangulation of document analysis and expert interviews conducted via video call or phone. We examined regulations, policy documents and press releases by the European, national and state levelsas well as reports by trade unions and regional border networks. In total, 28 experts were interviewed on the EU, national and regional levels, including policy-makers, civil servants as well as advisors from regional partnerships, employment authorities and trade unions (see list of interviews in Appendix I).

Empirical analysis
The following section analyses COVID-19 related disruptions and the re-establishment of the FMP across different governmental levels and in different German border regions. We find that transgovernmental networks were key for overcoming the initial restrictions to FMP and in avoiding similar disruptions during later stages of the pandemic. Moreover, since these transgovernmental networks operate at different levels, they also account for significant variation at the regional level. We find that long-established transgovernmental networks, mostly in German border regions with old EU member states and Switzerland, helped to mitigate the disruptions of FMP more quickly and pragmatically than in German Eastern border regions.
In order to appreciate the variation across policies and regions (see Table  1), we look first at how the crisis was managed in the areas of borders, health and welfare. This shows how vertical transgovernmental networks were essential in providing rapid, functional responses to the crisis and how they became more homophilous over time. Afterwards we focus on differential responses across German regions to explain patterns of variation linked to their different historical experiences and neighbours.
Vertical transgovernmental networks: key actors in coordinating policy responses to the crisis As highlighted above, intra-EU mobility was heavily restricted during the first wave of the pandemic and these restrictions resulted from uncoordinated, unilateral actions of EU member states. Still, despite lasting security concerns throughout the entire pandemic and despite limited EU competences to address these concerns at the supranational level, the FMP was re-established by summer 2020 and largely maintained during subsequent waves of the pandemic.
We find that transgovernmental coordination structures on varying administrative levels and within several institutional fora were crucial for the resilience of the FMP (cf. Boin & Rhinard, 2023). At the European level, Coreper became the central negotiation platform of the Council during the early phase, since, in contrast to the other decision-making fora, it did not stop meeting in person (Russack & Fenner, 2020, p. 7). At the beginning of March 2020, the Croatian Council Presidency fully activated the EU's integrated political crisis response mechanism (IPCR), a pre-existing arrangement chaired by the Council that coordinates, in an informal manner, EU institutions, member states and additional relevant actors such as the European Centre for Disease Prevention and Control (ECDC) (Interviews 26, 27, 28;Hollis, 2020;Russack & Fenner, 2020, p. 7). Another informal networkinternally referred to as 'COVID-19 Information Group -Home Affairs' (COM (2020) 3250 final, p. 4, 11) or just 'DG Home Group'was established by the Commission in the early days of the pandemic. Until summer 2020, it held weekly meetings together with various DGs, member state representatives and even private companies in order to share information and best practices Subnational -Regional governments (Regional Entry Regulations), regional and local administrations Regional directorates & agencies of BA (Interviews 26, 27). Since border controls were among the earliest responses to the pandemic, the Directorate-General for Migration and Home Affairs (Home) was the natural place to discuss such measures within the Commission. Under the Schengen Borders Code, member states enjoy considerable freedom to reintroduce border controls in case of unforeseeable events and public emergency (Article 22 of Regulation (EC) 2016/399). Therefore, in the first phase of the crisis, the Commission did not have the legal competences to force member states to re-open their internal borders. Given the uncertainty surrounding the pandemic, the Commission understood the reasons behind national border closures and, instead of using adversarial tools such as infringement actions, preferred to deal with potential breaches through administrative means (e.g., administrative letters, Interviews 26, 27). Given these limitations, the Commission sought to mitigate the obstacles to the freedom of movement of workers through non-binding guidelines (Interview 26), e.g., by proposing exemptions from travel restrictions for 'essential workers' (COM(2020) 115 final, p. 2; 2020/C 86 I/01, point 23 and 2020/C 102 I/03, point 2). However, the practical implementation of these measures was not binding for member states and mostly left to the Technical Committee on the Free Movement of Workers (2020/C 102 I/03, point 3), which is a statutory body formed by representatives of member states (Zelano, 2018, p. 109). Therefore, although many national measures contemplated exceptions for different groups of mobile citizens, they only followed the Commission's guidelines loosely (Interviews 17,24). For example, Germany exempted cross-border workers, persons transporting goods and citizens with 'important reasons' from border closures. In September 2020, the German Council Presidency made the IPCR the central organ of coordination (Interviews 26, 27). After some months in which coordination had taken place in several parallel fora, the German Council Presidency wished to concentrate their work in one single arrangement. For instance, apart from the 'DG Home Group', the Commission also gathered expertise through the Health Security Committee, which assembled health experts of member states and became a crucial institution to coordinate national responses (Deruelle & Engeli, 2021, p. 1376, as well as the Vaccine Steering Board. The IPCR seemed the most plausible institution to coordinate the COVID-19 pandemic, since response measures cut across different policy areas and the competences affected (mainly borders, public security, health and welfare) remained in the hands of member states (Interviews 26, 28).
When border restrictions were lifted and borders remained largely open after summer 2020, the Council, Commission and EU member states continued to rely on soft coordination (cf. Kassim, 2023). Even with incidence rates increasing again and new variants of the COVID-19 virus spreading from September 2020 onwards, they remained committed to the broad consensus that 'internal border checks must be a last resort'. 3 The Commission's Directorate-General for Justice and Consumers (DG JUST) prepared Council Recommendation 2020/1475 on a coordinated approach to the restriction of free movement in response to the COVID-19-pandemic. The recommendation introduced a regional colour-code that helped to visualise their epidemiological situation and guide member states' decisions when it came to requirements for persons from high-incidence areas to quarantine or test themselves after arrival (Eijken & Rijpma, 2021, p. 45).
Another example of the important role of transgovernmental networks (and their limits) in protecting the FMP regarded the coordination of social security. The purpose of Regulation 883/2004, which has its origins in European legislation from the 1950s and 1970s, is to establish clear national responsibilities for the social security of EU mobile workers in order to prevent discrimination and to facilitate free movement. Despite great legal uncertainty and complex regulatory problems triggered by the pandemic (ELA, 2021, p. 2), EU member states could draw on the longstanding expertise of the Administrative Commission on the Coordination of Social Security Systems established by Regulation 883/2004 and composed of civil servants from national welfare authorities .
Members of the Administrative Commission typically have known each other for years, interact on a regular basis and, soon after the outbreak of the pandemic, developed a shared understanding that this was not a situation of political, distributive conflict, but one that necessitated quick and flexible problem-solving (Interview 22). Under the leadership of the French and German delegations, the Administrative Commission issued in May 2020 an internal guidance note on the COVID-19 pandemic (AC 074/20). Essentially, the note identified common challenges and presented possible solutions on how to flexibly apply existing social coordination rules. Its non-binding character facilitated quick agreement and helped to disseminate best practices rather than obliging member states to agree on binding (legislative) solutions that might have raised difficult distributive conflicts (Interview 22). This solution was seen as extremely functional and, therefore, after May 2020, the Administrative Commission updated and prolonged the guidance note several times.
One major issue included in the guidance note concerned the applicable rules for cross-border workers who shifted to telework during the pandemic. According to Regulation 883/2004, cross-border workers who commute on a regular basis to their workplace are subject to the social security regulation of their country of employment. However, if they perform more than 25 percent of their work from their home country, the applicable legislation changes (Art. 11 (3)). Since border restrictions and lockdown measures forced many workers to switch to telework, there was significant legal uncertainty about the applicable social security rules (Interviews 7, 8). The Commission initially proposed a rather legalistic approach on the basis of Article 16 of the Regulation, which would have required all employers to request exceptions from the 25 percent rule (2020/C 102 I/03, point 8). To avoid unnecessary bureaucratic burdens, member state administrations agreed to simply suspend Article 11(3) for the duration of the pandemic (Interviews 8, 22). Eventually, the EU Commission abandoned its proposal and endorsed the pragmatic solution proposed by member states.
Still, the area of social coordination also shows the limits of informal transgovernmental networks when domestic rules are fundamentally incompatible. In the majority of EU states, quarantine obligations were subject to social security regulations and compensation for loss of income was paid by health insurers. By contrast, in Germany, these compensation payments were regulated in the federal infection protection law; hence, citizens were only eligible to receive compensation if quarantine was mandated by German health authorities (Interviews 13, 24, 25), making it impossible to receive payment if the quarantine was ordered by their home country (Task Force Frontaliers de la Grande Région 2.0, 2020). Solving this problem would require a convergence of domestic rules on compensation payments or the mutual recognition of quarantine ordinancesmeasures which could not be taken by an informal network of civil servants.
Finally, the EU Digital COVID Certificate (Green Pass) is another example that shows the importance of transgovernmental coordination for the resilience of the FMPin this case, horizontal coordination 'orchestrated' from the EU level. On 17 March 2021, the European Commission proposed EU Regulation 2021/953, which was adopted by the European Parliament and the Council on 14 June 2021. The Digital Green Pass facilitated free movement, as it offered a simple digital gateway through which national certificates could be verified. The EU's first-mover advantageproviding an international standard before any other country or private actorcertainly contributed to the quick uptake of the certificate by EU member states and beyond (Interview 26). 4 Importantly, EU rules mainly defined the technical framework, thus enabling the interoperationality of national certificates, but did not affect member states' competences to issue certificates for vaccinated, tested and recovered citizens and to regulate the rights and obligations of these groups (2021/953 (n 29); Interview 26). However, whenever member states waived restrictions to free movement for any of these groups, they were obliged to mutually recognise the certificates of other member states. Moreover, the Regulation has to be read in combination with the nonbinding Council recommendation 2022/107 of 25 January 2022, which allowed member states to coordinate their border measures in terms of citizens' individual vaccination, test or recovery status (2022/107, n14), instead of applying general restrictive measures depending on regional infection rates.
In sum, we see how vertical transgovernmental networks bringing EU and national actors together were essential in managing the early response to the COVID-19 crisis. Two trends are worth underlining: first, those networks that existed previously (Administrative Commission and IPCR) were more successful in becoming focal points, which shows that similar backgrounds and previous cooperation helps to provide a more efficient and coordinated response to the crisis. In contrast, the absence of a clear focal point in the area of health provided better chances for the Commission to orchestrate a common response at the EU level (cf. Brooks et al., 2023). Second, there are limitations to the work of transgovernmental networks, especially when solutions require stepping out of a coordination problem and into a distributive one.

Horizontal networks in border regions with old EU member states and Switzerland
In addition to these vertical transgovernmental coordination at the European and national levels, we also find that horizontal networks in border regions are crucial for the re-establishment and protection of the FMP. We see considerable regional variation in these horizontal networks leading to varying extents of resilience of the FMP. The federal German government installed border checks on 16 March 2020 at its southern borders and closed most crossings to France, Austria, Luxembourg, Switzerland and Denmark on 20 March 2020. 5 At the Dutch border, Germany installed health checks for travellers between 13 March and 15 June 2020 (Mission Opérationnelle Transfron talière, 2021a, pp. 46-47). To the local population, these initial border restrictions came as a shock 6 , since borders were considered a thing of the distant past, largely invisible or even unknown for the younger generation (Interviews 7,17,24). Their effects were felt particularly in border regions with high shares of mobile citizens, who travel regularly to neighbouring regions for leisure or shopping activities (European Commission, 2020, pp. 32-33).
However, the restrictions in regions bordering old EU member states and Switzerland were more limited in time and scope than on the Eastern border, often thanks to the work of long-established horizontal transgovernmental networks. First, the importance of cross-border work for the regions was acknowledged and worker mobility protected from the beginning. For example, since 17 March 2020, the entry regulation of Baden-Württemberg allowed cross-border workers to reach their place of employment and did not demand tests or self-isolation upon return (Interviews 7, 6, 20) 7 ; the French border restrictions also exempted cross-border commuters (Premier Ministre, 2020). Here, the distinction into essential and non-essential workers proposed by the Commission hardly played a role. In comparison, cross-border mobility for non-economic activities was constrained more severely. German travel restrictions exempted persons travelling for 'important reasons', although proving this was challenging in practice. In contrast, the Netherlands refrained from border controls, while Denmark defined the German state of Schleswig-Holstein as a border region with special status, resulting in very limited hurdles for citizens from that state (Interview 8). 8 In summer 2020, when the German federal government lifted border restrictions, decision-makers in Western border regions clearly positioned themselves against renewed border controls and travel restrictions. Even without rule-making competences, these horizontal networks helped improve the implementation of existing rules in a cross-border context and, whenever problems occured, they provided feedback to regional and federal policy-makers (Interviews 11, 24; Mission Opérationnelle Transfrontalière, 2021a, p. 110). Our interviewees repeatedly emphasised the importance of trusted partners and established links to peers across the border. For instance, the Upper Rhine Conference has served as a forum for transgovernmental coordination between German, French and Swiss authorities since the 1970s. In this context of high uncertainty, its health policy committee could build on 20 years of experience to intensify the frequency and scope of its meetings to almost every week and support local health authorities, e.g., to track infection chains across the border (Der Bevollmächtigte des Landes Baden-Württemberg beim Bund, 2021, pp. 9-10; Interview 17). Long-established regional cooperation between the Netherlands, Germany and Belgium helped to create a joint 'Task Force Corona' (Mission Opérationnelle Transfrontalière, 2021b, p. 14) in which regional and national officials reconciled their policies in an unbureaucratic fashion to avoid renewed unilateral border closures (Interview 18). 9 Another example of interregional cooperation is the Greater Region composed of parts of Luxembourg, Belgium, France and Germany. Its Economic and Social Committee is the only cross-border institution in which representatives of labour, social and professional organisations can develop shared regional strategies. 10 In 2019, the Treaty of Aachen on Franco-German cooperation institutionalised this arrangement further by creating a 'Cross-border cooperation committee'. After the outbreak of the pandemic, it set up an emergency committee to manage patient transfers and reduce travel obstacles of cross-border workers and local residents (Mission Opérationelle Transfrontalière, 2021a, p. 113-114). Cooperation at the German-Danish border also built on centuries-old traditions of politicians and administrators working together in the Regionskontor Sønderjylland-Schleswig (Interviews 1, 8). These transgovernmental regional networks often had clear learning effects. For instance, when infection rates started to rise again in October 2020 and France and Switzerland were classified as high-risk areas, state governments along the Western border adopted a '24-hour rule', allowing citizens to cross borders and return within one day without test or quarantine obligations. 11 These measures were considered of key importance to maintain daily cross-border interactions beyond worker mobility (Interviews 7,17). The 24-hour rule was proposed in a resolution of the Swiss Federal Council; subsequently, the Upper Rhine Conference and its political body, the Upper Rhine Council, adopted this idea (Der Bevollmächtigte des Landes Baden-Württemberg beim Bund, 2021, pp. 9-10) and advocated for its inclusion in the regional infection protection laws of German states, which was later taken up in the federal infection protection law (Interview 17).
The example of the Upper Rhine Valley is instructive in two additional respects. First, it demonstrates that effective transgovernmental coordination is not dependent on EU membership. Switzerland was involved in all discussions from the beginning and its approach was characterised by its partners as very pragmatic and proactive in mitigating obstacles to the FMP (Interviews 7, 17, 24). For instance, when infection rates were rising again in the region in January 2021, the Swiss government justified its decision not to introduce new test obligations as 'this would only harm Switzerland' (Interview 24). 12 Secondly, as with social coordination at the EU level, regional transgovernmental coordination reached its limits in case of incompatible rules of different member states. One such regulatory conflict concerned the double taxation of short-term allowances for workers employed in Germany but residing abroad. These allowances are, first, subjected to a fictional tax by the German Labour Office and, later, income tax is deducted by their state of residence (Interviews 5, 7, 24). While the Dutch authorities quickly agreed to exempt German short-term allowances from income taxes, this issue of double taxation resulted in a major loss of income for French cross-border workers (Interviews 7, 24). Instead of administrative coordination, litigation had to settle the conflict: a decision by the Federal Social Court of Germany decided that the German labour office must not deduct any tax in these cases (Interview 22). 13 In sum, we see how in border regions where there was long-term cooperation and solid structures, the use of horizontal networks helped to find practical and rapid solutions to the crisis. Those cases where no solutions could be found were generally related to Type I governance problems, namely the existence of incompatible national laws or the emergence of distributive rather than coordination problems.

Horizontal networks on border regions with new EU member states
In comparison, disruptions of the FMP were even more drastic at Germany's Eastern border to the Czech Republic and Poland in at least two respects: they lasted longer and they affected cross-border workers to a larger extent.
Horizontal transgovernmental coordination is less developed with Germany's Eastern neighbours, whichapart from specificities of local labour marketsexplains the difficulties to restore unhindered cross-border mobility throughout the pandemic.
Border controls and travel restrictions at German Eastern borders were not limited to spring 2020, but resurfaced in autumn 2020 and winter 2021, which led to the perception of one long period of border restrictions (Interview 12). The first closure was initiated overnight on 13 March 2020 by the Czech government, arguing that cross-border workers from Germany imported the disease to the country, although the numbers of infected people in border areas were not dramatically higher than in other regions (Interview 11;Novotny, 2022, p. 592). The Polish government followed up by closing the borders on 15 March 2020 (Mission Opérationelle Transfrontalière, 2021a, p. 65). Both closures were unexpected and led to scenes that 'resembled the 1990s when Poland was not part of the European Union' (Mission Opérationnelle Transfrontalière, 2021a, p. 60). Exceptions were made for workers in some critical occupations, yet more limited than those suggested by the Commission's guidelines (Mission Opérationelle Transfrontalière, 2021a, p. 76;Interviews 11, 12). Czech border closures remained effective until they were annulled by the Municipal Court in Prague at the end of April 2020 (Novotny, 2022, p. 593, 595), while Polish border closures lasted until June 2020 (Mission Opérationnelle Transfrontalière, 2021a, p. 65). Thereafter, the measures were adapted to the decreasing infection rates and crossborder workers could commute to work on a daily basis, but had to test once a month (Novotny, 2022, p. 593). In autumn 2020 and winter 2021, it was the German side that declared Poland and the Czech Republic as risk areas and introduced test and quarantine requirements as well as new police checks at the border (Interview 11). Most severely, from mid-February 2021 until mid-April 2021, Germany suspended free travel from the Czech Republic (and the Austrian region of Tyrol), which had been classified as a virus variant area. 14 Cross-border workers in German Eastern border regions also faced greater challenges due to the specifics of the local labour market. A large proportion of cross-border workers from Poland and the Czech Republic is employed in low-wage services requiring physical presence at the workplace (Interviews 11, 12, 15, 23;Haist & Novotny, 2023, p. 126). Telework was, thus, rarely an option. Moreover, inward migration is prevalent at the German Eastern borders, which might explain a weaker political position of cross-border workers in these regions. Yet, these local specificities cannot explain alone the greater difficulties to re-establish FMP at Germany's Eastern borders. Even if cross-border workers might be politically weak, their employers are not. Indeed, both Bavaria and Saxony are dependent on cross-border workers in key sectors, such as health, elderly care, logistics and industry, and hence lobbied for uninterrupted FMP from the early days of the pandemic (Haist & Novotny, 2023, p. 126).
More importantly, policies on both sides of Germany's Eastern borders showed limited awareness for the concerns of mobile citizens. Test and quarantine measures were hardly coordinated and adapted to the realities faced by cross-border workers, who were forced de facto to temporarily relocate to Germany or stay at home (risking losing their jobs). During the first wave of the pandemic, Czech cross-border workers had to quarantine for two weeks after their return, while Polish workers had to isolate themselves at home and to present a daily test. Only after lobbying from the Federation of the Polish Euroregions were the Polish restrictions slightly lightened, meaning that the compulsory quarantine was removed and tests were only demanded every two weeks (Mission Opérationnelle Transfrontalière, 2021a, p. 123). Similarly, during the second wave of the pandemic, the German states of Saxony and Bavaria used their regional competence in health policy to impose test and quarantine requirements for mobile citizens that went beyond the requirements in most other German regions and partly ignored the recommendation of the EU Commission on essential workers (Interview 11). 15 When the Bavarian test regime was challenged by the Administrative Court in November 2020, Bavaria lobbied for the inclusion of strict test and quarantine requirements at the federal German level in January 2021 (Interview 15) and both, Bavaria and Saxony advocated the introduction of border controls vis-à-vis Czech Republic in February 2021 (Interview 11). 16 Since telework was not an option in these sectors, questions of social security coordination were less of an issue, but access to short-term allowances or compensation payments during quarantine became major problems. First, the German Labour Authority (BA) initially considered that cross-border workers who were not actually available for the German labour market were not eligible for short-term allowances. It took until May 2020 for the German Ministry of Labour and Social Affairs to clarify that cross-border workers could nevertheless receive them (Interviews 11,12,15). Second, with strict quarantine rules for contact persons in Poland and the Czech Republic and the German system compensating only for loss of income when quarantine measures had been mandated by German health authorities (Interviews 11, 15, 25), many cross-border commuters faced a dilemma: they could either disregard quarantine measures by travelling to Germanythereby risking sanctions in their home countryor try to be sent on sick leave in order to continue receiving payments from their health insurance (Interview 23).
The difficulties to reestablish and maintain the FMP in these border regions compared to Germany's Western borders can be largely explained by the weak (or complete absence of) transgovernmental networks with Polish and Czech authorities. Since free movement was only fully realised after the end of the transitional provisions in 2011, support structures for cross-border workers were still young and rather limited. Apart from several Interregional Trade Union Councils 17 , only one EURES partnership was formed since 2007. 18 Moreover, the unitary character of Poland and the Czech Republic made it more difficult for regional actors to play a role in the management of the crisis, either by not being able to identify the right partners across the border or missing the right channels to raise concerns to the federal/state level. Various interview partners from the German-Polish and German-Czech borders reported that cross-border cooperation lacked sustainable structures and, where they existed, were mostly focused on individual, temporary projects (Interviews 11, 12, 15). Rather than being able to build on established structures, mobile citizens and activists had to create new networks to coordinate themselves and make their voices heard, e.g., in various Facebook groups and the new Czech Association for Cross-Border Workers (Novotny, 2022, p. 595). 19 Where cross-border networks already existed, they were often overwhelmed with individual requests for advice and support, and found it difficult to raise their issues to the political level (Interviews 12,15,23). A case study from the Czech-German Euroregion Elbe Labe prepared for the Commission Directorate-General for Regional and Urban Policy (DG Regio) summarises this argument concisely: There was an intense exchange between the Saxon and the Czech governments but there was little knowledge of the impact of the measures on border regions. Regional and local institutions were not consulted, nor were municipalities, and were not made part of the decisions (…) the Euroregion's main role was to provide advice: giving up-to-date information on the situation, which was continuously changing and needed to be made clear and accessible (e.g., through translation). However, the cross-border cooperation structures did not have real influence on political decisions. (Mission Opérationnelle Transfrontalière, 2021b, pp. 6-8)

Conclusion
This article explained how the reopening of borders and the reinstallation of the FMP was possible after the first wave of the pandemic in spring 2020. Our comparative case study of cross-border workers in German border regions used a multi-level governance approach to illustrate that the resilience of the FMP was not a distributive conflict between member states and EU institutions but a coordination problem affecting all member states. This coordination conflict was highly dependent on the interplay between three policy fields: borders, welfare and health as well as policy actors from European, national and regional governance levels. The closure of internal borders did not only cause obstacles to the cross-border mobility of workers but also triggered new questions surrounding the coordination of welfare and health systems in cross-border regions. New issues like the (forced) shift to telework, the availability and proportionality of tests and the financial compensations in cases of furlough and quarantine show the widespread consequences of managing a transborder symmetrical crisis like the COVID-19 pandemic.
In this multi-level setting, the resilience of FMP depended upon transgovernmental networks, which made it possible to find pragmatic and flexible solutions to the crisis. As expected from the theory, we saw that the observed variation between different policies and border regions can be explained by the strength of pre-existing coordinative transgovernmental structures, namely vertical and horizontal networks. Vertical networks were essential to coordinate an initial response to the COVID-crisis at the European level. In particular, long established networks in the policy field of borders and welfare (Administrative Commission and IPCR) showed their strength during the crisis, while the absence of cooperation infrastructure in the field of health, provided leeway for the Commission to install a common solution at the EU level. The study of the impact of COVID-19 on cross-border workers in German border regions demonstrated that long-established vertical cooperation was essential to agree on functional and swift solutions during a crisis. For instance, while some regions found coordinated ways to arrange test requirements and capacities, others struggled to ensure social rights (e.g., short-term [furlough] money) for cross-border workers. Nonetheless, the case study also revealed the limits of vertical and horizontal transgovernmental coordinationespecially when national laws were incompatible or a distributive conflict between cooperation partners emerged.
In contrast, the case study could not support the theoretical expectation that the political system (federal or unitary) played a decisive role in explaining cooperation. What it does show is that the reinstitution of FMP after the initial disruption proved more resilient for cross-border workers than for other types of mobile citizens. This is certainly important if we understand cross-border regions not just from an economic but also a social, cultural and identitarian perspective. The pandemic caused a shock particularly to those regions that had grown into transboundary living spaces and the disruptions were felt not just by those who could not go to work over the border but also by a broader set of social networks (families, friends, associations, etc.). While this might help to underline the importance of the EU and free movement in these regions, it also shows that these intangible concerns were initially disregarded and the economic logic of FMP prevailed.
The findings of our case study allow us to draw broader implications for the study of multi-level governance in times of crisis. First, the findings largely support the argument that the strength of transgovernmental networks is crucial for the resilience of multi-level systems when facing external shocks. They also confirm that variation in the political system cannot alone explain the existence or absence of cross-border cooperation and call for further analysis of transgovernmental networks. Second, they demonstrate that a multi-level governance approach is a useful tool to understand EU policy-making beyond its usual areas of application (e.g., cohesion, education, agriculture). In our case, it helps us better explain the complex responses in a core field of EU integration, namely FMP. Finally, studying COVID-19-related disruptions and the re-establishment of FMP from a multi-level governance perspective directed our attention to the demanding interplay of actors often needed to implement EU policies on the ground. It may, thus, serve as a reminder in times of crises not to focus exclusively on new advances in EU level integration, but also to trace the actual implementation and potential transformation of existing EU rules in practice.