Learning and reform in the aftermath of COVID-19: The Swedish experience of learning from a creeping crisis

Abstract

The purpose of this article is to use prior public policy and administration research on policy learning in relation to crises to assess the Swedish government's learning in response to COVID-19. To make sense of the learning processes at play, the pandemic is framed as a creeping crisis. The analysis is temporally organized according to conceptualizations of inter- and intracrisis learning and post crisis reform, and further focused by known challenges to crisis learning from previous research. The study underscores the importance of shedding light on challenges to crisis learning in order to be better prepared when future crises occur. More specifically, the analysis shows how internal disagreements on the objectives of the response strategy and polarisation on how to handle the crisis restrained the Swedish government's ability to learn from crisis experiences (intercrisis learning). Although instances of policy learning during the crisis (intracrisis learning) are observed, prerequisites for reform are limited due to interorganisational collaboration difficulties, politicisation of the crisis management initiative as a whole, and emerging challenges crowding out the agenda.

Introduction

The global spread of COVID-19 in the spring of 2020 was a creeping, ambiguous, and cross-boundary threat that led to a worldwide and protracted crisis. In January 2020, the pandemic was declared by the World Health Organization (WHO), and on 5 May 2023, the WHO Director General declared the pandemic over (UN 2023). Thus, the timing is apt for assessing government learning in response to crises and in response to COVID-19. Recently, moreover, scholars have called for research to better understand conditions promoting learning before and during crisis episodes (Steen and Rønningsbakk 2021, 158; Crow et al. 2023, 11; Lee et al. 2020), as well as more long-term reform in response to crises (Zaki, Pattyn and Wayenberg 2023). This article takes on such calls by assessing previous research on policy learning in relation to crises and by shedding light on important factors affecting governmental learning from crises in general and in relation to the Swedish government's ability to learn from COVID-19. The Swedish case is selected for scrutiny as it received global attention at the time of the height of the pandemic. For one, Sweden suffered badly during COVID-19, especially during the first wave from March/April to June of 2020 but also in the second wave, which began in October/November 2020. In total, 19,187 people died with COVID-19 in a country with a 10.5 million population (NBHW 2023). This can be compared to the neighbouring country Norway with 4,896 deaths from a population of 5.5 million (Zahl et al. 2024). The Swedish strategy of soft recommendations rather than strict obligations and relying on individual responsibility to take precaution on a voluntary basis, rather than forceful government lockdowns, was seen as maverick compared to most other states (Petridou 2020). Therefore, it is of interest to delve deeper into the question of how the Swedish government learned in relation to the COVID-19 pandemic.

This seemingly straightforward question requires some further clarification concerning the study's main theoretical constructs. First, we draw on Sabatier and Jenkins-Smiths’ work on policy-oriented learning, which is a key theoretical concept used in their Advocacy Coalitions Framework (ACF). In line with Sabatier and Jenkins-Smith, we define policy learning as “enduring alterations of thought or behavioural intentions that result from experience and which are concerned with the attainment or revision of the precepts of the belief system of individuals or of collectives” (Sabatier and Jenkins-Smith 1993 cited in Henry et al. 2022, 117). Second, in clarifying who learns, how they learn, and what they learn, it is important to underline that this paper presents a case study analysis of the Swedish government's learning before, during and after COVID-19 where prior public policy and public administration research on policy learning in relation to crises is applied to the Swedish government's learning from the pandemic. Government, in this case, relates to all actions taken in response to COVID-19 by the cabinet, the Government Offices and its ministries and the central public authorities that sort under the Government Offices. The latter are included as they are key instruments that the government has at its disposal to implement government policy. Zooming in on how and what the Swedish government in a broad sense learned, relates to questions of how to observe and measure learning. We focus on observed changes in beliefs that are indirectly measured, and reported changes in beliefs by policy actors through the observation of statements in news media or in public data (Henry et al. 2022, 119). Sabatier and Jenkins-Smith (1993) distinguished between deep core, policy core and secondary beliefs, where the former is highly resistant to change, while the latter two are more adaptable. In this study, we deal with the latter two learning modes altering policy strategies, positions and instrumental decisions rather than sociopolitical and fundamental values.

Third, any study on policy learning in relation to crises will be challenged by temporal issues and even more so when dealing with a case like COVID-19. As the pandemic lasted for roughly three years, there is a need to distinguish how observed policy learning relates to time and also how the concept of crisis relates to temporal aspects. In doing so, we adopt a creeping crisis frame to our study. We take on the definition of Boin, Ekengren and Rhinard (2020, 222) and relate to the creeping crisis as “a threat to widely shared societal values or life-sustaining systems that evolves over time and space, is foreshadowed by precursor events, subject to varying degrees of political and/or societal attention, and impartially or insufficiently addressed by authorities”. In keeping with the creeping crisis frame, we avoid treating the empirical case as one neatly unfolding uniformly over time. Time is, moreover, included into the equation through the concepts of intracrisis learning, intercrisis learning and reform (Moynihan 2008; 2009). We acknowledge that these learning modes are theoretical simplifications and that real-world crisis learning will at times spill over into several learning modes. However, we argue that these temporal crisis modes are helpful in making sense of the policy learning trajectories at play in the Swedish COVID-19 case.

Previous research on crises and policy learning

A crisis is often referred to in the literature as a situation where decision makers perceive a threat to fundamental values and norms amid considerable time pressure and extensive uncertainty about the situation, which in turn creates a perceived pressure on actors in decision-making positions to take action (Rosenthal, ‘tHart and Charles 1989; Boin et al. 2016; Lipscy 2020). As such, crises are unwanted phenomena deviating from the norm (Topper and Lagadec 2013). States and societies are increasingly challenged by crises defying geographical, administrative and disciplinary boundaries (Ansell, Boin and Keller 2010). Crises vary in duration and in pace, especially regarding development and termination. Crisis researchers have for long distinguished between fast burning and slow burning crises due to their pace of development and termination (‘tHart and Boin 2001). Lately, scholars have used the label creeping crisis to make sense of prolonged crises with long incubation times. Creeping crises linger right in front of policy makers. They transform from downplayed risk to acute threat and back, and thus defy the easily delineated emergency dominated crisis such as, for example, disasters or major accidents (Boin, Ekengren and Rhinard 2020), which dominates practitioners understanding of crises as well as theory development in the field. There are a few recent examples of how researchers have applied the creeping crisis frame to explain policy making and practice during COVID-19 (Boin, Ekengren and Rhinard 2020; Saetren et al. 2023; Zaki 2024). In line with these studies, we argue that COVID-19 came creeping to Europe in late 2019 and early 2020. Initially, politicians resorted to precursing events in China, Iran and northern Italy with wishful thinking and standard procedures. However, these impartial early responses did not limit widespread societal spread of the virus across Europe and globally (Swedish government 2021a, 205).

Policy learning induced by crises is a form of experiential learning. When experiences are grounded in a crisis, specific challenges come to the fore. Crisis learning is separated from knee-jerk reactive adaptation as the learning process is fuelled by individuals’ and collectives’ conscious and goal-oriented actions. Crisis learning then requires conscious deliberation and the linking of new ways of tackling policy problems and solutions to shortcomings that led to or exacerbated the crisis (Deverell 2010). The temporality and rarity of the experiences from which lessons are learned distinguishes crisis learning from other types of experiential learning (Carley and Harrald 1997). Unlike ‘ordinary’ experiential learning, crisis learning takes place under intense, compressed and temporary conditions in a context characterized by time pressure, uncertainty and conflicting values at stake (Stern 1997; Nolte and Martin 2021), thus becoming more challenging than learning from everyday activities.

Research on policy learning and crises has shown that stability is a more frequent outcome than post-crisis reform (Dekker and Hansén 2004; Bannink and Resodihardjo 2006) and that when governmental institutions learn after crises, they do so slowly and in a fragmented way (Boin and ‘tHart 2003). Scholars have also suggested conditions conducive to crisis learning. Birkland (2006) suggested increased attention on the political agenda, mobilization of various groups related to the issue, and solution-oriented and idea-driven discussions. Boin and 'tHart (2003) found an environment characterized by consensus agreements rather than by conflicted politicization to be beneficial. Research has also shown that when large-scale and structural change occurs after crises, it is not necessarily directly linked to the crisis. Usually proposed reform after crises have been discussed prior to the crisis (Quarantelli and Dynes 1997; Kingdon 2003; Peters, Pierre and Randma-Liiv 2011).

The creeping perspective on crises applies well to policy learning as learning in relation to creeping crises takes place in different temporal phases varying in intensity. A creeping crisis builds up over an extended period. It can have more or less regular outbursts, and crisis termination is a lengthy affair (Boin, Ekengren and Rhinard 2020) Scholars argue that policy learning becomes increasingly important in creeping crisis contexts marked by evolving interpretations over time, manifestation in fluctuating cycles of intensity and variations in societal and political contention and social compliance (Zaki 2024). Regarding different intensity phases of crises, Nolte and Martin (2021, 799) found that the initial ‘chaos’ of a crisis leaves little room for learning, while later stages provide more opportunities. Especially the hot phase of a creeping crisis then challenges organizations’ capacities to the extent that everyday organization and working methods do not suffice. When conceptualizing a crisis as creeping, it becomes vital to notice and measure policy learning over time. This entails not only after the crisis but also during the crisis. To truly understand the relationship between the creeping crisis and learning, then we need to broaden the perspective and take learning during the crisis response process into account as well as learning in the crisis aftermath. Moynihan's distinction between intercrisis and intracrisis learning can be helpful in this regard. Intercrisis learning is defined as “learning from one crisis to prepare for another”, while intracrisis learning refers to “learning that seeks to improve the response during a single crisis episode” (Moynihan 2009, 189). Intracrisis learning is more challenging than intercrisis learning because it is deeply affected by time pressure and uncertainty (Moynihan 2009). Both learning modes, however, refer to changes at the cognitive and behavioural levels. There are differing views on whether changes on the one level precede the other (cf. e.g., Crossan, Lane and White 1999 and Kamkhaji and Radaelli 2017). Drawing on Sabatier and Jenkins-Smith's policy-oriented learning definition, however, favours indetermination regarding how cognitive thought and behavioural intentions relate to each other temporally. This indeterminacy dovetails nicely with the idea of the creeping crisis and the value of not looking upon policy learning in a neatly and temporally compartmented way. We focus instead on changes in beliefs that are observed when they are reported by representatives of the government in a broad sense through the observation of statements in social or news media or in legislative data (Henry et al. 2022, 119).

So far, then, we have defined the key constructs of the study in the concepts of the creeping crisis and policy learning, which in turn is temporally divided into inter and intracrisis learning and reform. These concepts make out the foundations for the empirical analysis. Before we unpack the Swedish government's experiences of learning in relation to COVID-19, however, we present an exposé of previous research on crisis and learning, highlighting known challenges that will further focus the analysis.

Challenges of intercrisis learning, intracrisis learning and reform

This section presents a number of challenges that previous public policy and administration research has raised on the topic of policy learning in times of crisis. These occur temporally throughout the crisis process and are introduced according to the learning modes of intercrisis learning, intracrisis learning and reform (Moynihan 2008; 2009).

Institutionalizing lessons from a crisis by incorporating them into the institutional environment means that learning is integrated into preparedness and prevention processes. Such intercrisis learning occurs after or in between crises and thus becomes part of preparing for the next challenge (Moynihan 2009). One important part of the intercrisis learning process is to collectively reflect on passed experiences (Lee, Yeo and Na 2020). The objective evaluation or commission into how the crisis occurred and how it was managed is an important mechanism in doing so. To draft an objective crisis evaluation, officials need to overcome several obstacles. For instance, individuals with central crisis management tasks may have played a part in causing the crisis or their actions during the crisis may have exacerbated the outcome (Mitroff et al. 1989). Being responsible for a systemic breakdown rarely warrants open and impartial evaluation. Stern (1997), in his study on crises and learning, emphasized the problem of cultural bias in post-crisis evaluations. Cultural bias can originate from the idea that it is a sensitive matter to criticize one's own organization and colleagues. Collegiality is another form of bias that is difficult to avoid, especially in a small country like Sweden where many senior officials enjoy friendly relations (Isaksson 2015). Bias can also take the form of a prestige-oriented culture where organizational members refrain from criticizing each other. Collegiality and bias hinder learning and will, over time, erode capacity (Stern 1997). Self-evaluations are particularly sensitive to such bias (Roberts and Wernstedt 2019), which suggests that evaluators and investigators should not have been directly involved in the crisis management.

Actors who are expected to learn from the crisis may be subjected to harsh scrutiny by news media and the public, which may hinder learning. High media scrutiny is a contextual factor distinguishing crisis learning from ordinary experiential learning. There is an inherent contradiction here: Crisis learning requires reflection and an open environment where previously agreed upon norms and working ways can be questioned (Broekema 2018). But for such an environment to be maintained, learning agents require a work climate characterized by calm, mutual trust and an absence of open conflict (Boin and ‘tHart 2003; Sabatier and Jenkins-Smith 1993). This is seldom compatible with critical mass media crisis scrutiny. External scrutiny is a democratic necessity, but it may hamper crisis learning (Birkland 2006).

While the actual work of putting together evaluation committees or commissions in most cases belongs to the intercrisis learning phase, some of the obstacles feed back to intracrisis learning aspects, such as documentation and transparency. For instance, crisis management teams tend to be organized according to hierarchical structures or non-fixed conditions such as previous experiences or mere presence. The tasks of record keeping and documentation tend to be neglected in these situations due to stress, time pressure and sensitive decision-making contexts. Documenting decisions and actions in crises is vital to prepare answers to questions and critique in the postoperational phase (Curnin et al. 2023). Insufficient documentation hampers investigators, researchers, and others interested in exposing why the crisis occurred, and what was done to mitigate it. Previous research on Swedish governmental crisis management moreover shows that it has been hard to trace decision making, motivations and alternatives after national crises, which is problematic from a democratic and transparency perspective (Beckman, Olsson and Wockelberg 2003).

Leaders need to strike a balance between compliance and adaptivity for an apt crisis response (Janssen and van der Voort 2016). In this respect, intracrisis learning is key to understanding when and how a response strategy needs to be adapted (Janssen and van der Voort 2020). Furthermore, there is a potential danger in acting only in accordance with established routines and plans. In times of crisis, organizations need to go beyond policies, routines and working ways to reach appropriate outcomes (Borodzics 2004; Boin 2019). Plans can delimit adaptivity and creativity when those qualities are needed the most (Clarke 1999). Plans also need to be renewed in the light of new knowledge for an organization to be prepared when the next crisis occurs. This requires organizations to be agile enough to adapt to new challenges (Moon 2020).

A crisis is an undesirable event, but also a dynamic process cutting across policy areas, which can rarely be handled by a single organization. Thus, collaboration becomes a big challenge. Collaboration across organizational boundaries is required in the crisis decision making and learning processes (Nilsson and Eriksson 2008; Ansell and Ghash 2008; Nohrstedt et al. 2018). Organizations are advised to work together to solve challenges, which can create problems in communication between organizations. In addition, crises often spur new and unidentified stakeholders (Deverell and Olsson 2009). A crisis can thus lead to completely new constellations of actors and institutions that have not previously worked together, and this may impede collaborative learning.

Research on how actors work together during crises put forth contrasting perspectives. The concept of the ‘rally around the flag’ implies that groups come together especially in the acute phase of a crisis with an explicit antagonist (O’Neal and Bryan 1995; Boin et al. 2016). Creeping crises that stretch over longer time periods, however, will encounter a downturn in such rallying and spikes in contestation (Zaki 2024). Times following after the hot phase of the crisis can be characterized by politically motivated conflicts, commissions and evaluations aimed at investigating responsibility and allotting blame (Hood 2002; Resodihardjo 2020; Boin, ‘tHart and McConnell 2009). Shifting blame between actors is referred to in the literature as ‘blame games’ (Boin, ‘tHart and McConnell 2009; Hood 2011). Blame games shed light on individual mistakes and shortcomings. In so doing, they can distract attention from the implementation of more comprehensive lessons.

Reform in the aftermath of crises is the third and final temporal policy learning mode. Learning in the wake of a crisis is a prolonged process. Those who evaluate crises and crisis management in the crisis aftermath should be aware that even though they may have carried out an impartial evaluation, the learning process has only begun. Lessons observed must become lessons learned by dissemination and implementation (Birkland 2006). To achieve desired results, reports should explicitly propose specific lessons and be tailored for the needs of the politicians and senior officials that it addresses (Torres, Preskill and Piontek 2004). This will increase their potential and pave the way for reform. Unfortunately, institutional perseverance required to learn from crises is often lacking. Policy makers tend to be distracted in the aftermath of crises (Stern 1997), as they want to move beyond undesirable events as quickly as possible with their minds set on returning to normalcy (Boin et al. 2016). This means that crisis lessons can be overshadowed by emerging challenges or by everyday workloads.

Differing views on what should be done after a crisis and how it should be done, is another problem that can lead to loss of momentum in postcrisis reform processes. For instance, vague compromise solutions may be given priority when seemingly more powerful proposals become politically difficult to implement. Reform in the wake of the crisis tends to take place in conflictual political arenas marked by committee hearings and commissions (Resodihardjo 2020). Conflicts can also take the form of legal battles over insurance matters and compensation claims (Shrivastava 1987). In prolonged or long shadow crises (‘tHart and Boin 2001), such processes can be long standing and described as the ‘crisis after the crisis’. The fact that groups with opposing political views, ideals and values differ on what should be done and how to make political gains can thus hamper post-crisis reform (Dekker and Hansén 2004). In summing up, public policy and administration research suggest a number of challenges related to policy learning in times of crisis. Regarding intercrisis learning, previous research suggests prior crisis evaluations and commissions to be key mechanisms. These, in turn are affected by contextual conditions such as cultural bias at play, media scrutiny and the level of conflict. Intracrisis learning is also challenged by contestation and by for example, levels of transparency, adaptability and collaboration. Research on postcrisis reform put forth distraction and institutional perseverance as main challenges. It should also be noted that these theoretical concepts do not rigidly stay in one learning mode or phase. Rather, learning challenges tend to spill over into various conceptual constructs. This is especially true in a creeping crisis, which will have protracted incubation and termination phases and several outbursts and periods of relative calm (Boin, Ekengren and Rhinard 2020).

Method

Research Strategy

This study presents a case study of how the Swedish government learned in relation to the COVID-19 pandemic. The case study strategy is beneficial when contemporary or emerging phenomena that need to be understood in their own context are elucidated (Eisenhardt 1989; Yin 2003; Flyvbjerg 2006). Rather than producing generalized knowledge by assigning a correctly selected sample to a larger universe through statistical generalization, a case study can promote inductive identification of new propositions for further research (George and Bennett 2004).

Data sampling strategy

The literature review was based on articles from the database EBSCO Political Science Complete. The search included the keywords crisis AND learning OR learned and was limited to scholarly journals and books published since 1990. Older articles and books that were referred to in the initial sample were also added. The empirical parts of the analysis are based on official government data and media reports. Official documents were downloaded from the web sites of authorities with an explicit responsibility to respond to pandemics and other crises, including MSB the Swedish Civil Contingencies Agency, the Government Offices, the Public Health Agency (PHA), and the Board of Health and Welfare. Further, independent and reliable national media were used to corroborate official statements from authorities’ spokespersons. Media sources were located through the national database Retriever.

Operationalization

In order to unpack the question of how the Swedish government learned in relation to the COVID-19 pandemic, some additional clarification is needed. In keeping with Sabatier and Jenkins-Smiths’ understanding of policy-oriented learning, we see learning as a process leading to a policy outcome (Henry et al. 2022). The process, in turn, involves lasting changes in thoughts or behavioural intentions arising from experience, focusing on achieving or revising the principles of beliefs shared by the government (Sabatier and Jenkins-Smith 1993). As the study deals with the Swedish government's learning before, during and after COVID-19, we focus on actions taken by the cabinet, the Government Offices and its ministries, as well as the central public authorities subordinate to the Government Offices in response to COVID-19. The actual learning that took place is observed and measured as reported changes in beliefs by the relevant sets of policy actors through observation of statements in news media or in legislative data (Henry et al. 2022).

Analysis strategy

The analysis contextualizes and explains empirical episodes from the Swedish government's management of the pandemic supported by findings from previous public policy and administration literature on policy learning and crises. The analysis is further temporally structured by Moynihan's (2008; 2009) conceptualizations intercrisis learning, intracrisis learning and conditions for reform. In accordance to the theoretical framework then, empirical episodes concerning learning before, during and after the pandemic are included in the analysis.

Context: Prior experience and crisis reform

According to the Swedish administrative model, the government holds executive decision-making power and is held accountable by the parliament. The government makes collective decisions. It is supported in this endeavour by relatively small ministries tasked with planning and preparation and larger central authorities. Most of the state's operational tasks are handled by public albeit relatively independent authorities that are subordinate to the government as a collective, not to individual ministers (Petersson 2005). While some of the authorities are governed by a board, most are governed and led by a director general who can make decisions on account of the authority.

The national crisis management system is structured according to a decentralized approach departing from three crisis management principles: responsibility, subsidiarity and similarity. According to these principles, crises should be managed where they occur, according to structures similar to those operating in normal time. Every authority is mandated to deal with its own responsibilities in regular times as well as in times of crisis. There is no specific crisis decree or state of emergency suspending normal executive, legislative or judiciary arrangements (SEDU 2021).

The national arrangements for crisis preparedness are largely shaped by crisis experiences. Some examples of crises that affected the institutional design are the 1994 Estonia ferry disaster, the 1998 Gothenburg fire disaster, riots in connection with the 2001 EU presidency and the 2003 assassination of Foreign Minister Lindh. These events were occasions for central authorities responsible for national crisis management to sharpen their crisis management capabilities. Civil preparedness and crisis management capacity became increasingly prioritized after the fall of the Soviet Union and even more so after 9/11. The 2001 Risk and vulnerability commission advocated an improved holistic view of vulnerability and security, which laid the ground for a new policy area and for a new government authority responsible for coordinating national crisis management (Swedish government 2001). Hence, a decentralized system with relatively clear areas of responsibility was in place. At the turn of the year 2004/2005, two extensive crises occured. First on 26 December 2004 the Southeast Asian tsunami became a national crisis as many Swedes were on holiday in affected parts of Thailand. The government and key authorities came up with a tardy response. Second, southern Sweden was hit by hurricane Gudrun in January 2005, which put the spotlight on the lack of rural infrastructure preparedness. The crisis management was thoroughly evaluated by government commissions, and they unearthed significant systemic shortcomings. The weakest link was the role of the Government Offices in crises. The Tsunami Commission underscored that the Prime Minister had failed to ensure clarity in the Government Offices regarding responsibility for crisis decision-making at the top of the hierarchy, despite inquiries highlighting these shortcomings (Swedish government 2005). Although the commission delivered a candid report, political contention delayed implementation of the proposals for years. Subsequently, extensive reform was undertaken. After the turnover from a social democratic to a centre-right government in 2006, the inquiry ‘Crisis Management in the Government Offices’ was launched in the Prime Minister's Office, and in accordance with the Tsunami Commission's recommendations, the Crisis Management Office was established in 2008. In addition, several authorities with important roles in the crisis management system merged, and new ones (e.g., MSB) were created. At this time, epidemiological crises were prioritized in national crisis preparedness and in exercises (Swedish government 2021a, 155; 163).

When the social democratic-green party coalition regained power in 2014, the Prime Minister demoted the Government Office's Crisis Management Office, moving it from the Prime Minister's Office to the Ministry of Justice (Carlsson and Holmström 2015). This indicated that the government either did not want responsibility for crisis management, or that it did not understand the needs and expectations emanating from the constitutional fact that the government has overall and coordinating responsibility governing the state in times of normalcy and crisis. ¹ To sum up, the decentralized national crisis management system had improved with each crisis experience, albeit there were still weak spots, such as the unclear role of the central government in times of crisis.

Analysis: Policy learning from COVID-19 in the Swedish case

Below we analyse the Swedish government's learning before, during and after the pandemic. Empirical episodes are deliberated upon in relation to findings from previous research on crises and policy learning. The analysis is organized according to Moynihan's (2008; 2009) conceptualizations intercrisis and intracrisis learning and conditions for reform.

Intercrisis learning: Learning from previous crises to the next

What had the Swedish government and its authorities learned from previous crises at the time when the pandemic came creeping? In the years leading up to COVID-19, national civil protection and emergency services focused mostly on hard security issues including territorial defence, military threats, cyber security, migration, organized crime and terrorism and on traditional civil protection tasks such as wildfire preparedness (Swedish government 2017). Recent relative emergency preparedness successes were traditional and geographically delimited police operations such as the terrorist attack on a downtown Stockholm high street in 2017 and major rescue service missions such as the 2018 wildfires. There was thus a certain satisfaction within the administrative apparatus regarding crisis preparedness, but this did not mean that preparedness for blurry and ambiguous creeping crises was satisfactory.

In an administrative climate where a threat such as a pandemic competed with a plethora of threats, the voices of epidemiologists and virologists for increased pandemic preparedness had a hard time breaking through the noise, especially since the Swedish Institute for Communicable Disease Control was discontinued and merged into the PHA in 2014 with a broader mission of promoting public health (cf. Swedish government 2021a, 120). In the first phase of the pandemic, moreover, it became clear that intercrisis learning had been unsatisfactory and that national pandemic preparedness was inadequate. The lack of stocks of protective equipment, insufficient protection of the most vulnerable older adults in nursing homes, and far too limited test capacities, were areas identified as failures early on by the government's commission (Swedish government 2022).

Further, the basic tenets of national crisis preparedness and disease control were fragmented. Despite reforms suggested by previous crisis commissions, the role of the central government in times of crisis was still unclear. Another shortcoming was the lack of conditions for collaboration and accountability arrangements between key authorities. Responsibility for crisis preparedness in the area of public health was divided between the regional and the central level and between two different professional groups working with different logics and expert knowledge. On one side were the emergency services experts at MSB, and on the other side, the epidemiologists and additional medical professionals at the regional level and at the PHA. Data indicate that the two central agencies had difficulties in getting along prior to the pandemic. One evident example was the PHA's response to MSB's request for preparedness based on different scenarios due to events in Wuhan before the pandemic had reached Sweden. The PHA replied to MSB that “All measures are based on handling individual cases correctly and informing the outside world objectively so incorrect scenarios are not made up by ignorant authorities” (Larsson 2020).

In sum, the preparedness system was fragmented, and there was a lack of trustful relations between key players expected to lead in the event of a health crisis. As the creeping crisis appeared in full bloom, it became evident that preparedness for a pandemic was inadequate.

Intracrisis learning: Learning during the crisis

COVID-19 differed from previous outbreaks. Uncertainty and ambiguity were rife. For example, the virus did not affect humans as previous worldwide influenzas had. The first waves of COVID-19 did not sweep over all geographical areas laying down country after country. Instead, the virus initially infected people and geographical areas in clusters and therefore hit some regions, cities and neighbourhoods harder than others. Population density seemed to play a role, and how the disease control and crisis preparedness institutions of the geographical area reacted to the virus and what measures they put in place at critical points in time. Early evidence indicated that the virus could be suppressed with draconian measures that were seen as distant and unsuitable to Western democracies. A 76-day lockdown of Wuhan began in January 2020, which almost eradicated the virus in the city. Hence, Wuhan avoided extensive new waves of the virus (Björklund 2020). Such evidence, however, was not compatible with Swedish political culture. As state epidemiologist Anders Tegnell put it in his national radio talk in the summer of 2020: “This spring, the world went mad and country after country closed down. During the 2009 swine flu pandemic, nothing was closed” (Radio Sweden 2020). The intracrisis learning at the PHA was to stick to international agreements and national plans established before the pandemic, even after policy makers in neighbouring countries departed in different directions by imposing lockdowns and restrictive measures (Tegnell and Härgestam 2023).

Expert authorities put a lot of resources into transparency by crisis communication. Representatives appeared patiently at daily press conferences and interviews. On the other hand, transparency regarding how decisions were made, and what alternatives were raised at crucial stages of the crisis management was limited. In hindsight, it was not even possible to map out how the national COVID-19 strategy had been established. This led to disagreements about the actual purpose of the government's strategy, i.e., the very basis for the pandemic response (Anderberg 2020). Although the actual purpose of the pandemic strategy was unclear, some of the underlining principals were more transparent. The PHA motivated its regulations and measures by stressing that they had to be accepted by the general population and seen as sustainable over time in view of the expected duration of the pandemic (Swedish Government 2022, 10).

Admittedly, however, transparency was limited regarding how crisis management was organized at the PHA. Despite the critique and judging by the PHA's spokespersons, there was consensus in the executive crisis management team that the path chosen was the right one (SVT 2020). A frequently mentioned thesis in crisis management research is that teams working intensively together may develop overly conformist thinking. Group members can act according to illusions of invulnerability and unanimity and become overconfident in their own ability. Deliberate as well as subconscious peer pressure can then be placed on dissidents to change their views. Members of the conformist group avoid posing critical questions and end up making bad decisions (Janis 1982). One way of countering such groupthink is to hold second chance meetings where the collective takes a new critical look at the problem framing, alternative specifications and selected solution. The role of the devil's advocate, responsible for voicing critical perspectives, can be included in the crisis management team. This function is tasked to ask critical questions and call for alternatives and reconsideration of measures. In the case of managing COVID-19 in Sweden, however, the PHA instead turned inwards and rejected outside experts and critics (Delin et al. 2020).

On a global level, many states also turned inward to manage the pandemic individually and isolated from alliances, which may seem surprising as crisis research asserts that collaboration is essential in times of crisis (O’Leary and Vij 2012; Fisk, Good and Nelson 2019). In the Swedish case, obstacles emerged regarding strategic level inter-organizational collaboration. Although MSB and PHA representatives held many joint press conferences, there were differences in their understanding of the problem and its potential solutions. For instance, MSB's Director General in an email with foreign government representatives emphasized that: “MSB and I advocated faster and more comprehensive measures many times. But the Ministry of Social Affairs and their authorities chose a different path” (Delin and Carlsson 2021). Evidently then, MSB and PHA and their respective ministry had differing opinions on crisis response. Much of the disagreements regarded the value of precaution. State epidemiologist Anders Tegnell dismissed being guided by precaution, which is a guiding principle for MSB with its focus on rescue service and crisis preparedness. Instead, he claimed that: “To do more is not to follow the precautionary principle. Doing more can also mean doing more harm” (Sydsvenskan 2020). The issue of widespread use of face masks is the clearest example of differing views on precaution. During the crisis, the PHA consistently argued that there was a lack of evidence that use of face masks in society was effective, despite a growing number of research articles supporting the positive effect of widespread mask use (e.g., Howard 2021). Complete evidence was lacking, but most states came together in support of the benefits of face masks. Even the WHO appealed to Sweden to introduce face mask recommendations (Karlsmark 2020). Finally, the PHA partially reversed its view of the public's use of face masks by advocating use in public transport rush hours and in situations when social distance could not be maintained. At that time, however, the idea of the possible usefulness of face masks was severely undermined among the public.

The public debate about the pandemic response cannot be depicted as a rally around the flag. Rather, the pandemic divided the Swedish public debate into polarized camps where experts, politicians and the public held diametrically opposed views on how the crisis was handled. This turned into a typical politicized and mediatized struggle common in times of crisis. A potential consequence for the learning process in this regard was the government's difficulties in finding right competencies for its commission. The commission was initiated in the midst of the creeping crisis in June 2020, when the first wave was approaching its end (Swedish government 2022). According to media reports, it was a challenge to find medical expertise who had not been involved in or influenced by the debate during the first wave (Bjarnefors 2020). To avoid bias, the government refrained from appointing people who had been directly or indirectly involved in the management of the pandemic or in the public debate on the crisis response. Based on this strategy, the commission was established without consulting experts with strong views on national crisis management and disease control during the pandemic (Swedish government 2022). The commission thus lacked members with knowledge in key areas such as virology and infectious diseases.

The fact that the pandemic lasted so long, that it affected many areas, and that it varied in terms of intensity over time, prompted many opportunities for involved actors to learn lessons. One such opportunity was the period of relative calm in the late summer of 2020, which could have facilitated lessons learned from the first wave to the second. In the weeks and days before the second wave kicked in, the government eased several recommendations (concerning visits to nursing homes, audiences at public events and advice for older adults). Announcing softer recommendations shortly before the second wave, suggests that the PHA and the government once again underestimated the threat and that they were surprised by the progress of the virus in late October 2020. However, during the second wave in the fall of 2020 and in the beginning of 2021, the government took on a more active role in crisis decision making and went beyond the PHA's recommendations on several occasions (Lönegård 2020). The government also pushed through the temporary COVID act, which was passed on 8 January 2021 and revoked on 1 April 2022. This extraordinary legislation increased the government's authority over public gatherings, public events, shopping centres, public transport and air traffic (Government Bill 2020).

In summing up the intracrisis learning, the PHA staunchly stood by its initial and planned response. In the event of adaptation, the agency seemed to adapt grudgingly rather than as result of intracrisis learning. The government's increasingly active and visible role in the crisis management, on the other hand, suggests that the government took intracrisis lessons of the need for political crisis leadership to heart.

Learning in the crisis aftermath and conditions for reform

The pandemic led to a global state of emergency characterized by financial downturn in many industries. Aviation, tourism and travel, hospitality, and the culture and entertainment sectors were among the hardest hit. For the medical sector, the crisis left behind a staggering number of care cases that had to be served as medical staff was increasingly pressured by heavy workload (Swedish government 2022, 243).

The PHA's and the government's decision making and recommendations during the pandemic became highly politicized. The debate was linked to complex political issues such as privatization in the welfare sector, care conditions for older adults and individuals with chronic illnesses, just-in-time delivery and the state's responsibility for public safety and security. Moreover, the Swedish administrative system was criticized for lacking state of emergency laws and accountability mechanisms (Delin et al. 2020; Swedish Government 2022). The decentralized structure of the crisis management system and the disease control and prevention system also came under fire as they were criticized for being fragmented and unclear in terms of accountability (Swedish government 2021a, 148).

The government's COVID-19 commission raised specific critique to authorities. It found MSB to have an unclear and marginalized role during the pandemic and criticized the PHA for taking an overly exclusive lead in crisis management, while the government remained on the side lines. Further, the PHAs recommendations should have been formulated and communicated as clear guidelines rather than leaving room for personal interpretation. Regarding the lack of stockpiles for disaster medicine preparedness, the commission placed accountability on the regional level (Swedish government 2022, 650–653).

The commission called for a clearer system for political crisis management as well as a specific reform of the crisis management principles, by adding a precautionary principle (Ibid: 671). By this time, there had been a change of government as the social democratic government led by Andersson had been replaced by Kristersson's right-wing government following the results of the 2022 election.

Meanwhile, the new government struggled with several challenges that attracted media attention, such as spikes in violence tied to organized crime, immigrant integration, economic recession, climate change and not least territorial defence. Russia's increasingly aggressive foreign policy culminating in the full-scale military attack on Ukraine in February 2022, shifted the focus of the crisis preparedness policy subsystem even further toward territorial defence, military build-up and NATO accession (Stenberg and Lundborg 2024).

Thus, there are many distractions that could thwart learning processes in the aftermath of COVID-19. Whether the government and authorities have the perseverance to see reforms through and move from lessons observed to truly learned, is still a relatively open question.

Discussion

How could a developed, prosperous and security-conscious state like Sweden be so poorly prepared for the pandemic? One explanation is that medical healthcare issues were not prioritized in the crisis preparedness and security debate. There were many practical experiences in the medical field of pandemics. Examples included the Asian flu in the 1950s, and the 1968 Hong Kong flu. Lessons, however, had fallen into oblivion as the years had passed. Besides earlier pandemic experiences, there were also more recent ones such as SARS in 2002, the 2009 swine flu and Middle East Respiratory Syndrome (MERS) in 2012. Practical lessons from managing these pandemics were limited however, as SARS and MERS, both coronaviruses like COVID-19, did not spread actively in Sweden (Swedish government 2022, 632). Hence, these experiences did not provide crisis managers with first-hand knowledge or lessons that could be easily adapted to the new situation. Consequently, these experiences were not dominant in discussions on security strategies and preparedness in the years before the pandemic. In the national security strategy, for instance, the threat of an outbreak was seen as a challenge for hospitals and medical care, not for society at large (Swedish government 2017, 24). Conditions were worsened by the fragmented and decentralized system for disease control and prevention. Responsibility for keeping stockpiles of equipment needed during a pandemic, such as protective clothing, needles, respirators, and disposable masks rested on the regions and individual hospitals (Public Health Authority 2019; Swedish government 2022). These institutions, however, operate according to lean fiscal principles.

The case data indicate that there were challenges to the collaboration between two of the central government agencies in charge of the crisis response. If crisis managers have not built trust in each other and between organizations before the crisis, achieving results will be difficult (Ansell and Ghash 2008; Emerson, Nabatchi and Balogh 2012). These conditions thus affected the prospects for crisis learning in a negative way. In addition, the clash between MSB, an authority characterized by rescue services and the precautionary principle, and the PHA, where many decision makers are medical doctors who work exclusively evidence based was evident. Decision makers could receive guidance from previous cases, but decisions in creeping crises need to be made under extensive uncertainty and based on scenario thinking rather than on hard evidence (Shapiro and Bedi 2007). As all crises, the pandemic was rife with uncertainty and lack of evidence. Working exclusively evidence based thus became problematic during the pandemic.

As the case indicates, hard and restrictive measures were avoided with reference to plans, protocol and prior outbreaks, even though those outbreaks had little in common with COVID-19. The data, however, indicate that even though the government continued to avoid strict and obligatory measures in the second wave of the pandemic, the Prime Minister and his ministers had learned their lessons from the first wave. Thus, the government assumed a stronger role as crisis manager in the second wave, rather than shifting responsibility to expert authorities.

Case evidence of disagreements about the actual purpose of the national strategy, combined with missing documentation, meant that it was difficult for the government commission to reconstruct decision making processes in retrospectively (Swedish government 2022, 340). Lacking structures for transparency and record keeping in times of crisis thus again proved to be a challenge to learning. On that note, the government in April 2021 commissioned a report into the principle of public access to official documents and conditions for the media and the public to scrutinize authorities' management of COVID-19. However, that commission was not allowed to propose legislative changes (Swedish government 2021b), which indicates that the government did not in earnest prioritize crisis transparency reforms.

The PHA justified its regulations on the grounds of sustainability. An alternative approach could have been a more adaptive strategy, capable of evolving in response to prevailing conditions and emerging threats. An adaptive strategy entails the ability to adapt to changing conditions, which is considered an asset in times of crisis (Borodzics 2004; Boin and Lagadec 2000; Boin 2019; Ansell, Sørensen and Torfing 2021). If a decision does not reach intended results, it needs to be reanalyzed and adapted accordingly. The ability to adapt to new developments is beneficial for both crisis management and learning. When old knowledge is insufficient to solve the situation, actors need to be adaptive enough to look for new solutions. This is important as old norms can lead organizations into new crises (Turner 1976). In the case of the Swedish pandemic response, the tendency to reconsider previously prevailing views with an adaptive attitude was not a guiding principle for the PHA, most likely due to the authority's delimiting of outside experts and critical voices, and reliance on in-house competencies. This had effects on the potential for learning within the leadership. Outside expertise could have contributed to a broader understanding of the crisis and alternative coping strategies. The government on the other hand learned from the first and second wave of the pandemic as it adapted its strategies in time for the third wave and assumed more of a crisis leadership role, pushing through extraordinary legislation that increased the government's mandates instead of simply delegating decisions to the PHA.

Regarding more long-term reform processes in the aftermath of COVID-19, hopes were high for an agenda for change in social health policy. The pandemic exposed previously well-known problems, and discussions on these issues could be used as a springboard for policy entrepreneurs to launch their pet solutions (Kingdon 2003). This process was expected to shape domestic policy making in the long term. On the other hand, reform processes are affected by current events, and since the Russian full-scale invasion of Ukraine in February 2022, the issue of territorial defence, military support to Ukraine and the North Atlantic Treaty Organization (NATO) membership crowded out COVID-19 lessons learned and reform in the public debate. Somewhat counterintuitively, however, the Swedish NATO membership may also promote attention to COVID-19 lessons. NATO's 7th baseline requirement emphasizes the “Ability to deal with mass casualties and disruptive health crises: ensuring that civilian health systems can cope and that sufficient medical supplies are stocked and secure” (NATO 2023), which is similar to one of the main conclusions of the government commission regarding what was lacking in Swedish preparedness when COVID-19 escalated from a creeping to a full-blown crisis. As military and civil defence is high on the agenda, a NATO framing of issues of civil preparedness, disease control and mass care could become a viable way for policy entrepreneurs to push ahead reform.

Conclusion

The purpose of this paper was to use prior public policy and administration research on crisis and policy learning to assess the Swedish government's learning in response to COVID-19. To this end, the case of Swedish learning in relation to COVID-19 was analysed through a framework departing from the ACF definition of policy-oriented learning and the creeping crisis perspective. The analysis was further focused by challenges to crisis learning emphasized in previous research and structured according to the temporal divisions of intra crisis learning, intercrisis learning and reform in the aftermath of crises. The analysis put forth challenges to crisis learning such as distraction, bias, contention, accountability, transparency, adaptivity and perseverance. A conclusion from this discussion is that an environment characterized by calm and non-conflict where incidents, crises and decisions are documented, analysed and evaluated is beneficial for institutional crisis learning. Commissions play a key role in governmental learning from crises. On that note, fears that the commission's investigation following COVID-19 would be undermined by domestic and international politicization of Sweden's pandemic management and due to the absence of delegates knowledgeable in virology and infectious diseases, have not materialized. The government's commission suggested a number of potential and systemic reforms. However, new devastating events such as Russia's war in Ukraine have diverted attention from the reform agenda.

Applying the creeping crisis frame to the Swedish governments learning in relation to COVID-19 was helpful in explaining challenges, oversights and ambiguous policy outcomes. However, it also caused difficulties and variations when used in conjunction with Moynihan's temporal distinctions of crisis learning. While some of the challenges to policy learning noticed by previous literature fit neatly into the modes of intercrisis learning, intracrisis learning and postcrisis reform, the protracted and temporally challenging aspects of the creeping crisis defied the empirical fit of the theoretical learning modes. For instance, previous literature refers to crisis commissions and evaluations as intercrisis phenomena. Framed as a creeping crisis, the analysis of Swedish governmental learning during the pandemic, largely regarded the commission and its evaluation as a challenge of intracrisis learning. Commissions, however, may also have a long-term effect on postcrisis reform processes. This calls for further research to advance the creeping crisis frame as a tool for understanding contingencies that are hard to comprehend and compartmentalize.

As this study deals with lessons learned in a single case context, there are of course limitations to the potential of generalizing the findings or applying the results to other situations. To present more robust conclusions regarding governments’ learning from crises such as COVID-19, a larger sample of cases is needed. This study does not have the ambition to generalize its findings to other contexts. It should, however, be noted that many of the shortcomings of the Swedish management of COVID-19 were evident also in other countries. When viewed from a wider perspective, Sweden was not worse off than many other countries. For instance, studies show that Sweden, along with its Nordic neighbours, recorded among the lowest excess mortality rates in Europe in 2020–2022 (SCB 2023). According to Zahl et al. (2024) who compared COVID-19 mortality rates and policy measures in Sweden and Norway, 3,915 COVID-19 deaths in Sweden could have been prevented if Sweden had imlemented precautionary measures similar to those in Norway. The authors moreover questioned the justification of the Norwegian measures as they entailed huge societal and financial costs.

Notwithstanding this limitation, this study contributes to both scholarly and practical knowledge by applying theoretical concepts from public policy and administration literature on policy learning to illustrative examples of a national COVID-19 response and learning case. By examining the Swedish case, this study provides empirical findings that inform scholars and practictioners about how crises shape learning processes. This is noteworthy given that states, societies and organizations are increasingly challenged by crises, and as learning from crises is imperative for prudent crisis decision making, robust preparedness and effective future response.