For safety’s sake: the strategies of institutional entrepreneurs and bureaucratic reforms in Swedish crisis management, 2001–2009*

Abstract This study focuses on the bureaucratic reforms in Sweden which resulted in the creation of the Secretariat for Crisis Management and the Swedish Civil Contingencies Agency. We investigate the mechanisms that lead to divergent change through the critical juncture analytical approach. The study’s findings suggest that the bureaucratic reforms were the result of the critical juncture between 2001 and 2009, which included, inter alia, the release of a commission of inquiry report, a major political crisis and a national election. Moreover, we situate entrepreneurial agency in this analysis while we contribute to the theorization of institutional entrepreneurship by focusing on the implementation stage of institutional change. In order to overcome the institutional resistance stemming from an attempt to preserve the existing power structures, institutional entrepreneurs use the following three strategies: (i) the strategy of listening; (ii) the strategy of advertising early success and (iii) the strategy of picking up the phone.


Introduction
Crises, that is, jarring events that disrupt the way we understand the world and our place in it, are part of our everyday-ness -populating our twitter feeds, making headlines in print media and breaking news on popular news channels. The choice of events that constitute a crisis, albeit socially constructed (Bruck, 1992;Gupta, 2013;McBeth, Clemons, Husmann, Kusko, & Gaarden, 2013), often opens up space for change in the course of the political perspective of crisis management (Boin, Mcconnell, & 't Hart, 2008;'t Hart & Boin, 2001;Kingdon, 1984Kingdon, /2003Miles & Petridou, 2015). Change, if it occurs, is contingent on several factors, including the political context in which the crisis took place (see Schwartz & McConnell, 2008) and entrepreneurial agency (David, 2015;Hogan & Feeney, 2012;Kingdon 1984Kingdon /2003 In this article, we focus on entrepreneurial agency as we investigate the re-definition of crisis management policy in Sweden starting in the early 2000s, partly articulated through bureaucratic reforms. These signaled a fundamental change in the Swedish mindset around crisis management as for the first time we see enhanced cross-sectoral coordination of crisis management activities in a country where, unlike most, agency autonomy is paramount. It is worth noting here that the political crisis caused in Sweden by the tsunami in S.E. Asia was an extraordinary test for the Swedish system and has been typologized as an 'incomprehensible crisis' similar to the 9/11 attacks and Katrina (Boin et al., 2008;Brändström, Kuipers, & Daléus, 2008). These events are so unique that they open up an even broader space for agency than usual.
We unpack the mechanisms of change by mapping the critical juncture resulting in these bureaucratic reforms while at the same time situating entrepreneurial agency in the process of realizing and implementing institutional change. The critical juncture here is delineated between 2001 and 2009, comprising (i)  For the purposes of this study, institutional change is understood to be the establishment of formal organizations. The ideational shift in the thinking about crisis management that took place in Sweden was embedded in the new bureaucratic organizations and therefore they are the subject of our investigation. Also in this analysis, we narrow the focus of entrepreneurial agency to the strategies of institutional entrepreneurs. We identify three, and these are: the strategy of listening; the strategy of advertising early success; and the strategy of picking up the phone. The remainder of this paper is structured as follows: after the next section outlining the theoretical tenets underpinning this study, we present the methodology used to collect and analyze the data. Afterward we set the stage for the Swedish case study in which the ensuing analysis is situated. The paper is brought to a close with a set of concluding remarks.

Structure: critical junctures
The critical junctures framework within historical institutionalism is used to analyze a series of events, which when considered together, produce a certain result at a specific point in time by setting off processes of institutional or policy change (Donnelly & Hogan, 2012). Critical junctures are times of heightened uncertainty and contingency that result in the 'loosening of the constraints of structure' (Soifer, 2012(Soifer, , p. 1573 while maximizing actors' choices; thus, the past is a less salient determinant of future arrangements (Braun, 2015;Capoccia & Kelemen, 2007;Donnelly & Hogan, 2012;Hogan & Doyle, 2007;Kilinç, 2014;Mabee, 2011;Mahoney, 2002).
A critical juncture may be further unpacked to these analytical elements which render it 'critical' . First, antecedent conditions (the critical antecedent) though devoid of causal salience, set the stage for the factors operating within the critical juncture to eventually lead to divergence (Slater & Simmons, 2010;Soifer, 2012). Second, in the background of the critical antecedent, a set of permissive conditions emerge. These are identified as 'factors or conditions that change the underlying context to increase the causal power or contingency and thus the prospects for divergence' (Soifer, 2012(Soifer, , p. 1574, emphasis in the original). These conditions mark the window of opportunity (Kingdon 1984(Kingdon /2003. Finally, and once institutional constraints have loosened, a set of productive conditions shape the divergent outcome resulting from the critical juncture. Neither kind of causal condition in isolation is a sufficient cause of change; instead permissive and productive conditions should be seen as individually necessary and jointly sufficient (Soifer, 2012).
Though several studies have utilized critical juncture theory comparatively (see for example Collier & Collier, 1991), a number of recent studies have analyzed the factors which when considered together, form sufficient cause for change in single case studies such as the one examined in this paper (see, e.g. Consterdine & Hampshire, 2014;Kilinç, 2014).
Crises as components of critical junctures are contributing factors to the loosening of institutional rigidities and organizational interests (Cohen, Cuéllar, & Weingast, 2006;Kingdon 1984Kingdon /2003. In this paper, we follow the Cohen, Cuéllar and Weingast political definition of crisis as '(a) a circumstance perceived as an exogenous shock 1 sharply raising demand for policy changes in a particular domain, and (b) costly for politicians to ignore (2006, p. 707)' . Cohen et al. (2006) trace dramatic bureaucratic reforms in the US under the Bush administration that saw the creation a new Department, that of Homeland Security (DHS), in the aftermath of 9/11. Interestingly, though the authors' focus is on DHS, the authors also recount the creation of a structure within the White House National Security Council predating 9/11, tasked to coordinate issues of prevention and response regarding terrorist threats. This was a structure similar to the Secretariat investigated in this paper and also had its roots to the shift of the dominant discourse from mutually assured destruction to homeland security and from the threat of war to the threat of terrorism by non-state actors; all in a broader context of a growing -in size, scope and complexity -governmental apparatus (Cohen et al., 2006). Unlike the Swedish administrative state with independent agencies, the American political system allows for large agencies with sweeping powers and the authors succinctly tell the story of how the crisis of 9/11 was the opportunity for the extraordinary expansion of bureaucratic powers in a reform that saw 22 agencies subsumed under the DHS.
Therefore, crises may create the environment that enables agents of change to bring their ideas to the fore (Donnelly & Hogan, 2012;Kingdon, 1984Kingdon, /2003. Crises result in a more-than-usual attentive public the national leadership to take action, while at the same time exposing decision-makers to criticism thus potentially rendering inaction politically costly (Cohen et al., 2006;Hogan & Feeney, 2012).

Agency: entrepreneurship
Agents of change can act entrepreneurially if they strategically attempt to place divergent change on the agenda and realize it, partly through softening up policy communities and 1 in current social science discourse, a crisis is not so much a single point in time caused by a singular decision, but rather a process consisting of several (mostly endogenous) triggers leading to an existential threat (aradau & van Munster, 2011;Boin, 't Hart, Stern, & Sundelius, 2005).
larger publics (Kingdon, 1984(Kingdon, /2003. The process of softening up may involve actions such as speaking engagements, introducing and promoting new bills, submitting proposals, writing editorials or newsletters (Mintrom, 2000). Crucial to entrepreneurial behavior are the elements of creativity, purposefulness and perseverance: entrepreneurs must act purposefully and play the long game since policy communities and institutions are inertia bound and resistant to divergent change (Kingdon, 1984(Kingdon, /2003Meydani, 2009;Mintrom, 2000;Schneider & Teske, 1992;Schneider, Teske, & Mintrom, 1995;Sheingate, 2003). The work of Kingdon (1984Kingdon ( /2003 and Mintrom and colleagues (2000Mintrom and colleagues ( , 2013Mintrom and colleagues ( , 2015; Norman (2009), Mintrom andVergari (1996) center on policy entrepreneurs and especially on the initiation of new programs of policies rather than the implementation of them. In a parallel vein, a similar discussion regarding the limits and nature of the agentic capacity of individuals and change is echoed in theories of institutionalism (Petridou, 2017). Structures do not fully account for all complex political phenomena (Pierre, Peters, & Stoker, 2008, p. 234); a considerable degree of uncertainty depends on the actions of actors, who, in order to be classified as (institutional) entrepreneurs, must have 'an interest in particular institutional arrangements and who leverage resources to create new institutions or to transform existing ones' (DiMaggio, 1988;Hay, 2002;Maguire, Hardy, & Lawrence, 2004, p. 657). Moreover, institutional entrepreneurs are actors credited with carrying new ideas relating to specific institutional arrangements (Lowndes, 2005), and who purposefully utilize their resources to initiate and implement divergent institutional change (Battilana, Leca, & Boxenbaum, 2009;Garud, Hardy, & Maguire, 2007;Lawrence, Suddaby, & Leca, 2011).
However, in a departure from Battilana et al. (2009), we consider actors who implement change to be institutional entrepreneurs regardless of whether they initiated the change. The reason behind this theoretical decision is that in the case of bureaucratic reforms (as is the case in this study) such a requirement would by definition exclude the civil servants who are called upon to implement the institutional change and overcome the resistance that existing institutional settlements are likely to exert. This resistance is a direct result of shifting power relationships (Lowndes, 2005; see also Olsson, 2016), which civil servants have to navigate.

Methodology
We traced the process of institutional change by identifying the causal salience of the components of the critical juncture with the help mainly of elite interviews as well as document analysis. Data collected within the spatial and temporal boundaries of the institutional creation and reform instance under investigation are 'us [ed] [as] evidence from within the case to make inferences about causal explanations for [this] case' (Bennett & Checkel, 2014, p. 4), here specifically with the aim of theory testing (Beach & Pedersen, 2013). Critical junctures and the institutional softening exposed by crises are well documented in the literature, as is exceptional agency in the form of institutional entrepreneurship, providing the theoretical priors for this study.
Documents comprise commissions of inquiry reports; governmental reports; bills; the 2006 Moderate Party (M) party platform, and media (print) coverage. Additionally, and most importantly, we conducted 17 elite interviews between April 2015 and February 2016. Interviewees were identified partly through snowball sampling and chosen the basis of the role they played in the establishment of the Secretariat as well as the creation of MSB and included high-level bureaucrats and politicians. All interviewees are named in the Appendix 1 along with their relevant positions held during the years under investigation (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009). All were face-to-face interviews and lasted an average of 60 min. All but one were recorded and transcribed and the texts were systematically analyzed through coding using Atlas.ti. The vast majority of the interviews were conducted in Swedish; we translated the quotes used in this article ourselves. The use of multiple sources of data as well as the fact that both authors coded and analyzed material increases confidence in the validity of the coding scheme as well as the reliability of the process (Friese, 2014;Yin, 1994).
The theoretical priors mentioned earlier in this section served as a springboard for coding, based, initially, on a deductive logic. The output of the interviews was coded for themes, which comprised the main code families including agency; change; perception of crisis; ideological differences in crisis management philosophy underpinning the actions of the two major political coalitions; policy formulation; power and political support; entrepreneurial strategies, and articulation of (institutional) resistance. We revised the coding scheme during two additional rounds, during which we developed the analytical character of the codes as well as introduced inductive logic to the scheme. This resulted in the theoretical development of institutionally entrepreneurial strategies elaborated later in this paper. The coding process produced 60 codes in the eight main code families outlined above.
Theoretical priors were also used to construct hoop tests in order to increase confidence in our causal mechanism (Beach & Pedersen, 2013;van Evera, 1997). Here we must note that our ambition is not to explore causal mechanisms or examine interactions between macro and micro levels (see Bakir & Jarvis, 2017), but rather to reveal proposed mechanism understood as "intervening processes through which causes exert their effects" (Goertz & Mahoney, 2012). Hoop tests belong to a family of four empirical tests which facilitate the application of process tracing to causal inference. The classification of these tests is based on whether passing them provides a necessary and/or sufficient criterion for accepting the inference. Passing a hoop tests means that a hypothesis has to 'jump through a hoop' , however, accepting it does not, in and of itself, prove the inference. In other words, it provides a necessary, but not sufficient criterion. (Collier, 2011). To this end, we also employed counterfactual logic to provide an alternate explanation for the same outcome thus addressing equifinality, a common issue in process tracing (Bennett & Checkel, 2014;Goertz & Mahoney, 2012).

The Swedish Government structure and policy-making at the national level
Sweden's administration is divided into three main levels: central, regional and local. The Government Offices (Regeringskansliet) is the central administrative entity with staff assisting the Government (Regeringen) in policy preparation and governing. It comprises the Office of the Prime Minister (Statsrådsberedningen) and all the ministries. The Office of the Prime Minister's mandate is to lead and coordinate work in the Government Offices and thus holds a prominent place. The government and the Government Offices are two different organizations, though the prime minister is the director of the Government Offices. During a crisis, the government has the responsibility of national coordination and to lead the country. In its support to the government, the Government Offices have three distinct roles that in practice may bleed into one another: to be the government's support staff; (ii) to retain its structure as a bureaucracy; and (iii) to be the executive body responsible for Swedish subjects abroad (Lindgren, 2012). The entire office employs some 4600 people (Regeringskansliet, n.d.), which makes for very small ministries. Conversely, about 250,000 civil servants are employed in boards and more than 300 government agencies (for a detailed description and a diagram see Larsson & Bäck, 2008, p. 17; see also Sundström, 2016). These agencies perform the governmental work which in other countries is performed by the ministries. Large, fairly autonomous agencies, a small government office and the clear division between the two is the empirical manifestation of dualism -the politics/administration dichotomy -in Sweden (Hall, 2016).
One reason why Sweden's ministries are small is that the preparatory work in advance of a government bill is carried out by commissions of inquiry. Most of these commissions are ad hoc, appointed by the parliament and generally their members reflect the parties with seats in the parliament, though in two-thirds of them a civil servant of the relevant ministry is part of the commission as an expert or secretary. Therefore, the commissions can be seen as extensions of the Government Offices. There is a referral process after the completion of the report where the relevant ministries and agencies may submit comments. The process concludes with the government drafting a bill and submitting it to the parliament (Larsson & Bäck, 2008). The authors of the commission report have regular meetings and constant negotiations with the politicians who ordered the investigation. In practice, any conflicts regarding the contents of the report are teased out during that time. Petersson (2016) notes that increasingly commissions of inquiry become less independent, especially with the assignment of one special investigator with a support staff as opposed to a team of investigators.

Analysis
This section starts by setting up the stage for the critical juncture. We then discuss the permissive and productive conditions as well as the role of agency in terms of the strategies of the institutional entrepreneurs. In advance of the analysis and in order to guide the reader through the series of events, a timeline follows (Figure 1).

Critical antecedent: the drawback of the military, devolution and the need for coordination
The end of the cold war led to major changes in Swedish defense policy, which during the 1990s shifted from an operational defense force geared towards the protection of the state from invasion by foreign powers (from the East specifically), to a more technically advanced force focused on rapid action as well as providing support to international missions. The mandatory military service was rolled back with a concomitant process of professionalization, during which the military staff was cut by more than 90% and the vast majority of military regiments were shut down (SOU, 2001:41). This led to a sharp decline in resources that could be tapped should a crisis occur (Sparf, 2014). Similar measures were taken to restructure the civil defense; while previously civil defense was designed to ameliorate the effects of war, its new objective was societal protection in a broader sense including the security of critical societal functions as well as providing support to the military defense apparatus.
The shift in defense policy and the concurrent organizational changes aligned with a general restructuring of public agencies and authorities. Traditionally, the Swedish public sector has been large. However, following the economic slump of the 1980s, the country witnessed a neoliberal turn (Sundström, 2016). Authorities merged to form larger units and the state turned a wide range of public services into state companies. In some cases, complete sectors, such as electricity production, telephone market, railroad traffic, harbors, airports, schools, kindergartens, hospitals and care facilities have been deregulated (Hogan & Doyle, 2007;Kuhlmann & Wollmann, 2014). The organizational changes regarding crisis preparedness -from military defense to civil authorities, county boards and municipalitiesstarted in the 1980s.
Concomitant to the shift from military defense to new bureaucratic arrangements, was the realization that the dividing line between very serious accidents and extraordinary events was increasingly blurry and diffused. The central government saw that there was a need for coordination at the national level in order to provide the support the county boards and the municipalities would need in order to manage societal crises (Statskontoret, 2012). The politics and legal issues of horizontal coordination in Sweden were (and still are) persistent questions. In 2003, the then DG of the Swedish Emergency Management Agency (SEMA), Ann-Louise Eksborg, explored the question of who is in charge in times of crisis and concluded that decision-making by a single entity is unconstitutional and instead, especially at the operative level, coordination is crucial (Eksborg, 2003).

Permissive conditions: the Swedish Commission on Vulnerability and Security and the tsunami crisis
It was against this backdrop that the of Swedish crisis management reform took place, leading to a devolution of responsibility to local administrations and an integrated approach opening up for a wide range of new actors -public as well as private. In 1999, a commission of inquiry (The Swedish Commission on Vulnerability and Security) was set up by the Minister of Defense to update the structure of civil defense and emergency preparedness planning, which admittedly, was 'set up for war and had not been used in 250 years' (R07). We regard the release of the commission's report as a permissive condition facilitating the ensuing change. This is because it marked the awareness of the Swedish government that the existing structures built around the concept of civil defense did not match the kinds of crises Sweden would likely face in a post-cold war era.
This commission proposed the three principles which the parliament voted into law and are in effect today: (i) the principle of responsibility, under which entities responsible for an activity during normal times retain this responsibility in crisis or war; (ii) the principle of parity, under which authorities retain their structure and location in crisis or war, and (iii) the principle of proximity, under which crises should be handled at the lowest possible level of government (SOU, 2001:41). In line with the Swedish consensual sensibilities, this crisis management paradigm is not based on unified command, but rather focused on coordination and collaboration. According to a critical voice (R03), the Swedish system is cumbersome because of the large number of commissions of inquiry, investigations and evaluations partly designed to achieve consensus, while lacking in decisive action.
To strengthen coordination at all levels of government, the commission went further in suggesting organizational changes at the national level. More specifically, it proposed that a national crisis management unit be placed in the Government Offices (Regeringskansliet), and that a new 'planning' agency be established with the mandate to provide strategic planning; coordinate research and development efforts in the field of crisis management; provide some intelligence, as well as support the crisis management unit in the Government Offices (SOU, 2001:41).
In 2002, the planning agency was established as Krisberedskapsmyndigheten (KBM). It is worth noting that the official name of the agency in English is 'Swedish Emergency Management Agency' (SEMA), but the Swedish says 'Crisis Preparedness Agency' . Either way, there was a mismatch between the name of the agency and its mandate, which did not include an operative role during a crisis (R17; Fichtelius, 2007). This made it impossible to justify the agency to Swedish public after the media maelstrom of the tsunami and thus made its existence unsustainable. We will return to this later.
Conversely, the proposal to establish a crisis management unit in the Government Offices was not followed by then PM Göran Persson's social democratic government -at least not then. This was a conscious choice rooted in the fundamental attitude among the social democrats that the prime minister must be protected at all costs. The most serious crisis would be one that might threaten to topple the government, and 'the closer to the prime minister [the Secretariat for Crisis Management] is, the more dangerous politically it is' (R02). Indeed '[f]rom the Social Democratic side they were like, you cannot have such a body in the Prime Minister's Office, crisis management is risky business, I mean, you can lose your job, you can lose your government' (R01). At the same time, [t]he prime minister wanted actually to control to lead [crisis management], but did not want to have a dedicated bureaucracy [for it] [...] Rather he wanted to choose when to intervene and that was the case even regarding the media. Not in the least in order to control who within the government took charge of different situations. He was not really inclined to give the power to one agency to be singlehandedly responsible for crisis management. (R16) In December of 2004, a crisis did occur -the kind that engenders 10-year anniversary pieces in major international news outlets. About 250,000 people died (SOU, 2005:104) and many more were left homeless after the tsunami that hit 14 countries in Southeast Asia. The overwhelming majority of these casualties were local inhabitants but there was a large number of foreign nationals vacationing since Thailand and Sri Lanka are popular Christmas holiday destinations for many and especially for Scandinavians. At the time of the catastrophe, about 4000 Norwegians, 3800 Finns and 4500 Danes were in the area (SOU, 2005:104). Among the Scandinavian countries, Sweden was hit the worst. About 30,000 Swedes (the equivalent of a mid-sized Swedish town) were vacationing mostly in Thailand and Sri Lanka, and figures from November, 2005, put the missing at 543, the dead at 525, and the injured at about 1500. (Brändström et al., 2008;SOU, 2005:104).
Not having somebody in charge seems to have backfired for the Persson government, giving the appearance that the government was caught sleeping, that it was unresponsive and disorganized (Buller & Carlson, 2005;Nilsson, 2005;Wolodarski, 2005). The political crisis and the blame games that ensued have been examined elsewhere (see for example, Brändström, 2016) and are beyond the scope of this paper.
Crucial for our argument, though, is how the tsunami crisis exposed institutional shortfalls in the Swedish crisis management thinking, which can be summarized thus: [a]nd all the work we had done up until the tsunami we had to throw away because it was a completely new situation then, completely new things to deal with, shortfalls in the society's way to handle the crisis, mainly within the Government Offices and the coordination between the Government Offices and the authorities … before the tsunami you thought you could plan for everything, pretty much. (R04) All of our respondents, regardless of their position as a bureaucrat or politician, the latter even regardless of the party they belong to, expressed the same sentiment: there was no mechanism in place to deal with a crisis that was other than foreign invasion; there was no protocol to activate in a situation like this and the fact that the tsunami happened during the Christmas holidays when nobody is at the office in Sweden only made matters worse.
The commission of inquiry (Disaster Commission) set up in the aftermath of the crisis suggested, inter alia what the 2001 commission on Vulnerability and Security had suggested four years earlier: a crisis management unit to coordinate among governmental agencies and provide intelligence to the prime minister, placed directly under him/her in the Prime Minister's Office (SOU, 2005:104). Only then did the Persson government establish a unit called 'Unit for Preparedness and Analysis' (Enheten för Beredskap och Analys -EBA), which did not have much clout and 'was an embryo that I think patched a leak and was a bargaining chip to be able to say "hey look, we are doing something"' (R07). The office did not have access to the political leadership and did not hold daily or any other kind of briefing (R13); it was seen as a symbolic and rather cosmetic act without substance.

Productive condition: the election
The timing of the tsunami crisis and its handling by the Social Democratic government partly resulted in the center-right Alliance ascent into power after the 2006 election. The crisis exposed the operative inadequacy of KBM and the need of an agency that would have such an operative mandate during extraordinary events. The change of government came at a time when the discussions regarding a bureaucratic reorganization had been going on for some time; the sentiment was it was evident that something had to be done (R08). The Moderate Party (the largest party in the Alliance) had signaled already in their party platform prior to the election that they would tackle safety and security issues (Moderata Samlingspartiet, 2006) and when they came into power they used a basic strategy to show they were taking care of a social problem by creating (and dismantling) formal organizations (Jacobsson & Sundström, 2016).
By all accounts the decision to create the secretariat had been made and the argument was foreclosed by the time a commission of inquiry was set up to investigate crisis management in the Government Offices. All of our respondents said that this was what the Alliance, in fact what the newly elected PM Reinfeldt wanted -for the Secretariat to be under the Prime Minister's Office, to have resources for a 24-h operation arrangement and to serve as the cross-sectoral crisis coordinator in the Government Offices support the government with intelligence, education and exercises. The commission of inquiry was tasked to figure out the logistics rather than deliberate whether there should be a secretariat to begin with. As mentioned earlier in this paper, commissions of inquiry have been less independent during the recent years and that the trend is to assign one person (the special investigator) to write the report with the help with a supporting team, which was the case here (Salomonson, 2007).
Concomitantly, MSB was created by the merging of the State Rescue Services, the Swedish Emergency Management Agency and the National Board for Psychological Defense. This was a discussion the origins of which had predated the tsunami crisis. It was part of the adjustment of a bureaucratic structural system created for war -a crisis in sharp contrast to accidents or other peace time such crises -and the realization that the world has changed and bureaucracy has to be integrated enough to handle such peacetime crises in an integrated fashion. Our respondents viewed the creation of MSB as the result of a slow process that was going to happen regardless. However, the tsunami crisis and the subsequent election hastened the process and also determined the current mandate of the agency as geared towards society at large and as a support agency to the Secretariat. As one of our respondents put it, 'everything everybody was talking about for the last 15-20 years happened within the space of eight months' (R10).

The role of agency
Our in-depth research of the Swedish case reveals a nuanced picture of the role of agency in times of crisis. Agency was important in the institutionalization of the secretariat, but its establishment was fairly straightforward according to all our informants. This is partly due to the fact that the crisis management structures did not meet current needs but also reflects the power of the prime minister in the Swedish system and the rationalistic way policy is prepared through commissions of inquiry. Though our analysis revealed actors who felt that a change in crisis management policy was necessary through a coordinating bureaucratic apparatus of some form, policy entrepreneurship through the persistent softening of the policy community and the larger publics was not evidenced.
What is more, the organizational change articulated in the establishment of the secretariat under the prime minister's office was perhaps more incremental than it might appear at first glance. The idea had been suggested in commissions of inquiry in 2001 and 2005 and it was only due to a deniability reflex (a political reflex perhaps stemming from being in power for the majority of elections cycles post 1945) of the Social Democrats that kept the office as far away from the prime minister as possible and prevented it from being realized until 2006. Even when the early version of the secretariat (EBA) was established in 2006, it lacked muscle. Instead, and keeping with the Swedish dualistic system favoring large agencies, the Social Democrats were more keen to address the crisis management issue by conferring power to an agency. A governmental proposal submitted to the parliament (Riksdagen) 2005, calls for the creation of an agency tasked specifically to lead (vs. simply coordinate) in times of crisis. In practical terms, this one agency would inter alia have cross-sectoral decision-making authority (Prop, 2005/06:133). Such a solution was counter to the Swedish political reality since, unlike in most countries, agencies are fairly autonomous and they make virtually all decisions regarding their internal organization and work (Ehn, 2016).
These measures became moot after the election as the Alliance seized the political capital it gained from the election. It scrapped EBA and immediately appointed Christina Salomonson as a special investigator to explore not so much the establishment of a secretariat for crisis management under the Prime Minister's Office, but the logistics necessary. She was a well-liked politically neutral public servant with good reputation and long experience in the agencies that would be affected by the secretariat. It was understood that the process would be swift and sustained communication between the investigating team and the government during the writing of the report ensured a result everybody could agree on. The report (Salomonson, 2007) and the ensuing proposition by the government (Prop, 2007/08:92)  The Secretariat was established in an uncharacteristically swift fashion with all the resources specified in Salomonson's report (2007). 'It was like an open door' (R10) and the secretariat had full and explicit political support -a 'big hand on the back' (R01). That bid hand on the back, was HG Wessberg, the state secretary for the prime minister. Such explicit political support was needed if the reform was to be implemented that is, if the secretariat was to function as a cross-sectoral crisis management coordination unit in the Government Offices providing direct support to the prime minister. There was marked resistance by the governmental departments (ministries), and especially the Department of Foreign Affairs, the Defense Department and the Justice Department, all of which considered the Secretariat to be stepping on their toes by performing duties that were in their purview. They were keen to protect their turf by maintaining the existing power relationships.
It was then at this stage where agentic action became crucial for transformative change that is, the institutionalization of the new bureaucratic unit. The entrepreneurial actors in this instance were the then state secretary of the prime minister HG Wessberg, the first Head of Department Christina Salomonson and the first Deputy Head Lars Hedström. The strategies they used are different to the strategies used by policy entrepreneurs at the agenda setting stage, where the aim is to bring the issue along with the possible solution to it to the foreground. Here, the aim was to delineate the boundaries of the new bureaucratic unit and defend them against the existing units by playing the power game just right. More specifically and as Battilana, Leca and Boxenbaum put it, institutional entrepreneurs had 'to justify the divergence from taken-for-granted practices and frame the vision for that change in a way that enables others, despite its unfamiliar nature, to understand and endorse it ' (2009, p. 81). We arrived at the specificities of these strategies inductively, though as mentioned elsewhere in this paper, the coding process started deductively, as a theory-testing exercise. By the third round of coding on Atlas.ti, three strategies, complementary to the extant theory of institutional entrepreneurship, emerged. They are outlined below.

The strategy of listening
The first strategy used by the institutional entrepreneurs was the strategy of listening. Instead of the new unit throwing its weight around and thus disrupting further the existing power structure, the institutional entrepreneurs had 'big ears' (R01) in order to be responsive to the needs of political and bureaucratic actors as information consumers. What kind of information is relevant, how much and when? The ability to listen and provide non-threatening support in terms of information to the different ministries was a deciding factor towards gaining legitimacy, as the following quote suggests: '… [t]hen [the ministries] thought 'ah, ok, maybe this [piece of information] is good, we can use it' . So if the Secretariat got the opportunity to support [the various ministries], then it got help, then it got accepted' (R02). This strategy engages in dialog with, and in fact complements, the work of Battilana et al. (2009). The authors develop the strategy of 'motivational framing' as the providing other actors with 'compelling reasons ' (2009, p. 80) on why the new arrangement should be supported. We argue that 'talking' strategies have limitations and that successful entrepreneurs must have the ability to listen. Listening to the needs, wishes, intentions, and complaints of other actors can shape the provision of the 'compelling reasons ' Battilana et al. (2009, p. 80) why the divergent change ought to be institutionalized thus gaining the much-needed legitimacy to do so.

The strategy of advertising early success
Second, the secretariat took advantage of early successes, which it treated as 'educational' crises (R07; R13). On the one hand, communicating a timely and decisive handling of a crisis increased the confidence of the ministries to the Secretariat as a competent crisis management unit. Later contingencies qualifying the early success do not seem to detract from the salience of this strategy. For example, a decision to vaccinate the entire Swedish population against the swine flu in 2009 was communicated as successful crisis management even though later the vaccine was linked to instances of narcolepsy in young people. This strategy is similar to framing and the use of narratives that not only institutional (Battilana et al., 2009), but also policy entrepreneurs (Meydani, 2015;Mintrom, 2000;Schneider et al., 1995) employ in order to mobilize allies and convince them that the change is worth supporting. These were narratives of success and had the symbolic aim to convert actors resistant to the secretariat.
On the other hand, skillful crisis management contributed to the gaining of legitimacy in more practical terms as well. It brought the ministries within the Government Offices together and it was a way for people to interact, get to know each other and work together. These events lowered the threshold of cooperation by increasing the level of trust and decreasing the level of suspicion.

The strategy of picking up the phone
Third, and perhaps most importantly, the institutional entrepreneurs used their personal connections to shore up the legitimacy, access and ease of communication between the secretariat and the political leadership. This is not to be underestimated in a small country, where politics-by-picking-up-the-phone is the norm. People know each other and this is something that we discovered as well during our data collection: the crisis management field is small inside an already small Swedish state (see Jacobsson & Sundström, 2016 for a discussion on the latter). The overt political support, the 'big hand on the back' (R01), was partly the ability to pick up the phone and get things done thus blurring the dichotomy between politics and administration. This strategy fits nicely with the strategy of 'forming communication channels' when it comes to governing the state put forth by Jacobsson and Sundström (2016). Despite the dualism characterizing Sweden's administrative state (Hall, 2016) or perhaps because of it, informal networks between public officials and politicians become even more salient in the praxis of governance.
The strategy of picking up the phone nicely provides a nuance to Battilana, Leca and Boxenbaum's strategies of mobilizing allies and resources. Actors with formal authority are in a position to use their clout for entrepreneurial purposes. They are also able to tap social capital resources through their central position in informal networks. When asked about the importance of personal relationships, one of our respondents (R13) said that during Reinfeldt's second term, when the original players had left, the secretariat, lost some of its importance. He stressed the salience of the individual and their focus on bridging the silos of the Swedish administrative state through informal communication channels. He also pointed out that politicians and bureaucrats must navigate conflict situations through brokerage practices because the most importance purpose of public actors is to solve problems.

Hoop tests and counterfactual logic
To shore up the causal logic of our argument, we conducted a series of hoop tests (Collier, 2011;Mahoney, 2012;van Evera, 1997), with a view to showing that each of the permissive and productive conditions were necessary for different aspects of the divergence and we prove them using existing knowledge from extant literature (Mahoney, 2012). The utilization of hoop tests is applicable in this case, because as mentioned elsewhere in this paper -though not formally articulated as hypotheses, permissive and productive conditions are individually necessary and collectively sufficient. Taken cumulatively, these hoop tests support our causal inference by showing the mechanism that lead to the bureaucratic reforms.
• The 2001 commission of inquiry was a necessary condition for the bureaucratic reforms in terms of a qualitative change in how crisis management is done. We know from the literature that convening a commission of inquiry is a common way for the Swedish government to prepare policy recommendations (see for example Larsson & Bäck, 2008). • The crisis in the aftermath of the S.E. Asian tsunami was a necessary condition for the bureaucratic reforms in terms of the substantive nature and the mandate of the new bureaucracies. It is well established in the literature that crises loosen institutional rigidities and expose flaws in policies (see e.g. Birkland, 1998;Dror, 1993). Here, it exposed a lack of intelligence capacity and coordination at the Government Offices level as well as the lack of an agency with operative capacity. The question that was raised at the agency level was not necessarily whether there would be a new agency, but instead how much power it should have. • The change of government after the election was a necessary condition for the bureaucratic reforms in terms of timing. Jacobsson and Sundström (2016) explain how the creation of formal organizations and/or the dismantling of others is a common way for Swedish Governments to address social problems. The creation of two bureaucracies, one by dismantling three others, happened immediately after the center-right Alliance came to power.
A further question that emerges, then, is whether this would have happened if the permissive and productive conditions were not in place and specifically if the crisis had not taken place and the election results were different or had not taken place at that time. We address equifinality by applying counterfactual logic. We know from our elite interviews (for example R13; R16) that the Social Democrats in general and Göran Persson in particular absolutely did not want the secretariat too close to the prime minister. We thus claim that a version of the secretariat might have been established or EBA might have continued as it did, but the result would not have been a fully resourced, 24-h operation unit with direct access to the PM. In an interview to Erik Fichtelius, Göran Persson clearly states that his government did not agree with a center for crisis management at the Government Offices and that this role should be handled in the existing agency-level administrative structure (Fichtelius, 2007). Our counterfactual logic is supported by hindsight knowledge. One of the first things the Social Democratic coalition government did upon ascending to power again in 2014 was to move the Secretariat in the Justice Department under the interior minister.
Having said this and although the secretariat and MSB are two legs of the same policy sector, MSB would probably have been established with the merger of the three agencies for these reasons: (i) the discussion had started much earlier; (ii) the social democrats believe in large agencies and (iii) it was less politically sensitive. The question would then center on the scope of the agency and how much power it might have.

Conclusions
In this paper, we set out to understand the shift in crisis management policy and practice that occurred in Sweden during the 00s. This was evidenced by bureaucratic reforms resulting in the establishment of two bureaucratic organizations -the Secretariat for Crisis Management and the Swedish Civil Contingencies Agency. The creation of the new bureaucracies is explained through the analysis of the critical juncture between 2001 and 2009. The critical antecedent of the post-cold war era and the demilitarization of Sweden set the stage for this critical juncture. A commission report released in 2001 and the political crisis brought on by the tsunami which hit S.E. Asia in 2004 constituted the permissive conditions for divergent change, whereas the elections of 2006 were the productive condition acting as a catalyst for change. We further shored up the causal salience of our argument by the use of counterfactual logic and hoop tests demonstrating that each of these conditions was individually necessary and collectively sufficient, thus concluding that the bureaucratic reforms were the direct result of the events comprising the critical juncture. The creation of the Secretariat for Crisis Management and the MSB is further nuanced in the sense that the process for the latter had been ongoing for some time; the events of the critical juncture hastened it. However, the secretariat would not have been placed directly under the prime minister and would have had a different scope had it not been for the historical contingencies that led to its creation.
Furthermore, this research contributes to the theorization of institutional entrepreneurship, by shedding light to the implementation of institutional change, which tends to be neglected in the literature. More specifically, our findings complement the existing theorization of institutional entrepreneurship by demonstrating the value of listening as well as talking. Entrepreneurs who listen are able to provide non-threatening support to actors concerned about shifting power structures. Conversely, advertising early success has symbolic value in gaining legitimacy as well as lubricating the working relationships among public officials as well as between public officials and politicians. Finally, in the strategy of picking-up-the-phone, this research confirms the importance of personal relationships and the value of networks in navigating structural arrangements.
In conclusion, further research of comparative character would have the possibility to refine the results of this study. It would be fruitful, for example, to comparatively examine the implementation of bureaucratic reforms in terms of the strategies of institutional entrepreneurs. Alternatively, further research may test the strategies outlined in this paper in another bureaucratic reform in the Swedish context with a view to enhance the understanding of entrepreneurial agency in the Swedish administrative state.