When the mountain broke: disaster governance in Sierra Leone

Purpose – WhenamajorlandslideandfloodsdevastatedFreetown,SierraLeonehadjustovercometheEbola crisis, which had left its mark on socio-political relations between different disaster response actors. With international disaster response frameworks increasingly shifting to local ownership, the national government was expected to assume a coordinating role. However, in “ post-conflict ” settings such as Sierra Leone, intra-state and state – society relations are continuously being renegotiated. This study aimed to uncover the complexities of state-led disaster response in hybrid governance setting at national and community levels in the response to the 2017 landslide and floods. Design/methodology/approach – DuringthefourmonthsoffieldworkinFreetownin2017,semi-structured interviews and focus group discussions with various state, aid and societal actors were conducted. Findings – Thefindingsshowthataresponsetopolicybuildingontheideaofauniformstateresponsedidnot take into account intra-state power politics or the complexity of Sierra Leone ’ s hybrid governance. Practical implications – This paper argues for a more nuanced debate in humanitarian governance and practice on the localisation of aid in post-conflict and fragile settings. Originality/value – The study ’ s findings contribute to the literature on the disaster – conflict nexus, identifying paradoxes of localised disaster response in an environment with strong national – local tensions. The study highlights intra-local state dynamics that are usually overlooked but have a great impact on the legitimacy of different state authorities in disaster response.


Introduction
Worldwide, 318 disasters were recorded in 2017, affecting 122 countries and 96 million people (CRED and IRSS, 2018). One of these disasters had not attracted much attention but had impacted many lives. On 14 August 2017, Sugarloaf Mountain in Sierra Leone "broke". The ensuing landslide and floods swept through a densely populated area in Freetown, killing over 1,000 people and destroying hundreds of houses within minutes. The Sugarloaf Mountain disaster hit a country still impacted by the civil war that ended in 2002 and recovering from the Ebola epidemic that lasted until 2016.
Disaster response is generally organised in a top-down manner, but the international disaster governance literature has questioned the reliability of authorities and formalised response organisations (Comfort, 2007;Norris et al., 2008), increasingly emphasising that disaster response should involve multiple actors. The internationally recognised preferred response model proposed in the Sendai Framework for Action (UNISDR, 2015) views the state as the central coordinating body, working closely with civil society, communities and other non-state actors in disaster response and disaster risk reduction. This model, which is generally reflected in national disaster policies, is anchored on a well-functioning state; the present article asks how this plays out in fragile institutional environments such as Sierra Leone.
In countries with a recent history of conflict, state institutions often lack the required resources and capacities to respond to disasters. Ongoing institutional changes and sociopolitical tensions complicate their legitimacy to coordinate disaster response. Relations between different layers and domains of the state are in flux and are impacted by state-society relations. Furthermore, post-conflict states are usually subject to a strong international presence engaging in programmes for state building, peace building and recovery.
A key premise of the present study is that disaster response models are translated and socially negotiated in practice. Although different parties may agree on roles and responsibilities, how these work out in practice is negotiated, potentially with highly divergent outcomes. Therefore, it is important to use empirical case studies to explore how disaster response evolves, seeking potential patterns and broader lessons. With this aim, this article presents a detailed analysis of the response to the landslide and floods in Sierra Leone. As the case unfolded, our focus was drawn to the central importance of intra-state competition and the issue of vertical linkages in how disaster response is translated from the national to the local. These issues underlay many mundane contestations over aspects of the response, such as the registration process, the introduction of cash relief, item distribution and the screening of internally displaced persons (IDPs), to determine camp eligibility.
Our study thus aimed to understand how state, aid and societal actors negotiated the terms of a state-led response in Sierra Leone after the 2017 landslide and how this affected the legitimacy and institutional capacity of a multilayered state.
Sierra Leone's triple "posts" in perspective Sierra Leone is considered a postcolonial, post-conflict and post-disaster state, reflecting key periods during which intra-state and state-society relations have been challenged and redefined. These three "posts" shaped Sierra Leone and continue to affect the state of the country and its multilayered governance.
After gaining independence in 1961, Sierra Leone's postcolonial state embodied many remnants of colonial relations. The country's hybrid governance structure comprises continued interaction between the formal state, paramount chiefs, ceremonial chiefs and other community stakeholders. Authority is gained through a mix of state and traditional sources that have become mutually constitutive, but this also resulted in a governance culture of distrust and resentment (Albrecht and Moe, 2015;Jackson, 2007;Keen, 2005). In addition to formal recognition of their positions, the legitimacy of the paramount chiefs in the provinces and the ceremonial chiefs in Freetown comes from their traditional roles, which are partly rooted in strong patron-client politics and which took on particular importance under colonial governance (Harrell-Bond et al., 1978) when the chiefs functioned as gatekeepers between communities and the state (Cooper, 2002 in Anderson andBeresford, 2016). The political and social control of the state and traditional leaders marginalised young people against the backdrop of an economic crisis, leading them to join the rebel forces in the ensuing civil war (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002) to fight the patrimonial structures (Peters, 2011;Tom, 2014).
The period after the civil war redefined power relations within the state. To facilitate peaceful transitions, with international support, institutional reforms and reconciliation processes were enacted. Regarding disasters, of particular relevance are the Office of National Security (ONS); the Ministry of Social Welfare, Gender and Children's Affairs; and the National Commission for Social Action (NaCSA), all of which were established after the civil war. State-building processes revolved around the decentralisation of the state, where scholarly attention is directed to the contentious relationship between paramount chiefs and other formal state structures (Albrecht, 2017;Clayton et al., 2015;Jackson, 2007;Keen, 2005). However, much less is known about the roles and relations of ceremonial chiefscustomary leaders in the Western Area including Freetown. Generally, contention continued after the civil war, and young people, supported by the international human rights discourse, gained space to negotiate with and challenge traditional leaders, questioning their legitimacy (Tom, 2014).
The Ebola crisis (2014-2016) had both negative and positive impacts on Sierra Leonean institutions and state-society relations. Positive influences were seen in the organisation of disaster management structures; however, because of the scale and the nature of the disaster, the newly established National Ebola Response Centre took over the coordination and control of these structures from the ONS and the ministries. Further, tensions between society and the state were evident as the population ignored government messages on Ebola, perceptions of corruption persisted and confidence was eroded by the militarisation of the response (Anderson and Beresford, 2016;Lind and Ndebe, 2015). Chiefs played a pivotal role in the Ebola response on a community-level (Wilkinson and Fairhead, 2017), but this role has sometimes been contested by youth questioning their authority (Marcis et al., 2019).
When the landslide and floods struck in 2017, a relatively well-established response structure remained in place from the Ebola crisis. Nevertheless, the roles and relations in this structure continued to be fluid, and relations of trust and accountability were still uncertain.

Post-conflict states and international disaster response
Although national states traditionally play a major role in disaster response, in conflict-affected and fragile states, this often becomes primarily the domain of international humanitarian action. International actors and the national state are often seen to operate in isolation from each other (Cunningham, 2018). In reality, there is constant interaction between the two governance systems, although humanitarian agencies often seem to take control in disaster response (Barnett, 2011;Donini, 2012), undermining the state's authority. One challenge of post-conflict disaster governance is therefore how to break the pattern that evolved during conflict, where international actors dominate disaster response and tend to bypass the state.
The Sendai Framework and national-level disaster risk reduction frameworks and laws give prominence to the state in disaster response. However, post-conflict settings create many challenges for state-centred disaster response (Harrowell and € Ozerdem, 2018;Melis, 2018). Despite international policy's focus on states, in practice, states are often bypassed by humanitarian donors. In the 1970s and 1980s, international disaster aid usually consisted of budget support to the government, but from the 1990s onwards, international organisations increasingly took over disaster management (Harvey, 2013, p. 153). In 2017, only 2.5% of international humanitarian funding was channelled through national governments (GHA, 2018, p. 51).
The Sendai Framework mentions "states" 37 times, "government(s)" 33 times and "authorities" 17 times. Although it centres on the state, it focuses on the role of national governments, generally failing to differentiate between different layers and parts of the state. The state does not equal a government. A state is made up of different institutions, and leadership and authorities can also be found outside the government (Miliband, 2009). To some extent, this is recognised in the Sendai Framework's inclusive view of governance, which includes national and local authorities and communities (UNISDR, 2015, p. 13), but this disaster governance policy model seems to assume that governments form the centre of a network of disaster response with clear roles and responsibilities.
In contrast to this idea, ethnographic studies in post-conflict and postcolonial settings have revealed a far more complex situation than a uniform central state system, finding that the state comprises different entities and hybrid institutions (Boege et al., 2009), competing for authority and legitimacy at different governance levels. At national level, state entities are part of post-conflict state-building and the accompanying institutional changes (Brinkerhoff, 2005). At local level, different modes of local governance coexist, particularly in states where traditional authorities function as mediators between the state and society (Menkhaus, 2007;Olivier de Sardan, 2011). In patrimonial societies such as Sierra Leone, traditional structures occupy both formal and informal governance roles and are reshaped to accommodate modern state structures (Albrecht, 2017;Albrecht and Moe, 2015). Rather than the central state's will simply being translated through decentralised bodies, the state is also shaped from below through interaction between the central state and local authorities (Titeca and de Herdt, 2011). Disconnections between parts of the state may be more pronounced in post-conflict settings than in more stable areas. This study sought to understand disaster response governance at, and between, the national and community levels.

Methodology
This article is based on fieldwork conducted by the first author in Sierra Leone from September 2017 to January 2018. As the official response to 2017 landslide ended on 15 December 2017, a substantial part of the fieldwork consisted of real-time research of the response. This allowed great access to current information and made it possible to follow-up on leads regarding movements by responders and discussions in the IDP camps.
Data were collected through interviews, focus group discussions, participant observation and attendance at a "lessons learned" workshop organised by the ONS and the International Organisation for Migration. The semi-structured approach, multiple methods and variety of participants allowed for the triangulation of data. In total, 88 semi-structured interviews were conducted and audio-recorded with 37 aid actors (9 national NGOs, 14 International Organisations [IOs] [e.g. donors, United Nations [UN], 14 international non-governmental organisations [INGOs]); 32 community actors (27 IDPs, 2 civil society organisations, 3 volunteers); and 19 state actors (9 ONS representatives, 4 chiefs, 6 representatives from the NaCSA, the Ministry of Social Welfare, the military, the Ministry of Energy and Water Resources and the Environmental Protection Agency). These interviews followed the structure four 15-participant focus group discussions with community stakeholders, whereby a timeline of the response was created and the participatory tool of ranking and scoring was employed.
A grounded thematic analysis was conducted using NVivo qualitative data analysis software which centred on themes of actor relations, challenges, discourses and response practices. Participants were categorised by type of responder (state, non-state and community) to allow for multiple perspectives to emerge and inter-actor relations to be further analysed.
A research assistant living in an IDP camp provided translations for the community interviews. The communities of Mortormeh, Gbangbayila, Kamayama and Pentagon as well as the Juba and Old School camps were visited frequently to collect data. The research assistant's status as a flood-affected person from one of the communities risked introducing bias but also provided better access to the affected communities and, through close collaboration, increased our contextual understanding. In May 2018, a follow-up visit was made by both authors to validate the findings with research participants.
The state meets disaster: connecting the national and local The landslide and floods in Sierra Leone occurred in the capital city of Freetown; as a result, the response was structured in a relatively simple way, with a national, centralised response and the implementation of relief and recovery efforts in the affected communities.
National state-level governance: internal competition and international support As the disaster unfolded, all departments and ministries relevant to the response were called to meet in the situation room at the ONS. The sectoral response pillars that had been active during the Ebola response, with dual governmental and international agency leadership mirroring the UN cluster system, were reactivated. However, despite the discourse of unity, competing interests and power politics within the state and in collaborations with non-state actors revealed how different authorities within the state negotiated for position and power over and through the response.
The ONS has the mandate to coordinate emergency responses. As an interviewed national state official asserted, "It's like a sharp knife running through a ripe banana", meaning that the roles were clearly defined (GOV10 12.01.2018 int.) [1]. However, tensions between the different roles of the ministries and state institutions continued to play a role. One national state official explained the sources of these tensions: "The issue is that of mandates. Even though institutions may have the responsibility, for example that of social welfare; in a normal situation, it's their responsibility to seek the welfare of all citizens, but when it becomes an emergency, the responsibility of coordinating all of this rests with the ONS. [. . .] so we also have an issue of institutions not wanting to surrender some of their [power], subjecting themselves to be coordinated and given instructions" (GOV10 12.01.2018 int.).
To resolve this, "the president had to intervene and admonish every ministry that the responsibility of coordinating the emergency lies within the Office of National Security. Nobody should go there and give counter instructions" (GOV10 12.01.2018 int.).
In terms of mandates and capacity, confusion and contention arose, especially between the ONS, the Ministry of Social Welfare and the NaCSA. In the initial days after the landslide, these institutions all began to register affected households, with all three considering this to be within their mandates. The NaCSA was later informed, with the authority of the president, that the Ministry of Social Welfare was in charge of the registration. The tasks of each institution then became more defined.
However, the legitimacy of the Ministry of Social Welfare was still questioned, especially because they were thought to be incapable of the task, as is seen in the following comments from a national state official: "With all due respect to their commitment, but the expertise was not there for us to get the kind of data that we wanted. So it affected the whole intervention" (GOV9 10.01.2018 int.). And another national state official insisted, "We cann't accept them anymore. [The Ministry of] Social Welfare cannot handle registration in the future. They cannot. We can do it together" (GOV8 18.12.2017 int.).
Conversely, the Ministry of Social Welfare felt that the ONS, as the leader in disaster issues, was trying to control everything and stepping on their turf. A national state official explained, "There has been some amount of beef between the two institutions" (GOV7 17.12.2017 int.), referring to conflict between them. As a symbol of protest, the Minister of Social Welfare did not attend many coordination meetings, and the ONS, as a coordinating body, was unable to enforce its coordination efforts in the different ministries.
Although tensions were highest between the ONS and the Ministry of Social Welfare, challenges in intra-state coordination were also visible between different sectoral response pillars. Pillar leads from different ministries were supposed to remain in the ONS situation room to facilitate all parties signing off on requisitions. However, as a national state official recounted, some pillar leads stayed in their own ministries, which delayed the process (GOV4 07.11.2017 int.). Not all pillar leads were present during meetings or they would arrive late. Overall, there was significant reluctance to accept coordination by another institution within the state.
The ONS was the paramount body coordinating the disaster response. This required the temporary suspension of the mandates of other ministries, but these ministries' responses showed that they did not trust the process and feared that the disaster response would lead to a permanent loss of their mandates. A narrative of capacity was used by both institutions to discredit the other and strengthen their own authority. Tensions were also built into the system because, although the ONS had authority, they lacked resources, which remained within the ministries.
The disaster response structure was further complicated because of roles allotted to the military and to a fiduciary agency, BDO. The military were in charge of logistics, but before items could be taken out of storage, they had to be registered and signed off by both the ONS and BDO. BDO was hired at the beginning of the response to prevent a repetition of the corruption practices prevalent during the Ebola response. The intra-state bureaucracy and hierarchical systems led to delays in the release of aid items. By the end of the response, only a fraction of the donations had been used. Besides the friction this arrangement caused for the logistics pillar, which was in charge of the storage but did not have authority to move the items there, it also affected the ONS managers of IDP camps, who had to deal with the bureaucracy.
To complicate matters further, a hierarchical decision-making and implementation system was developed. The National Platform for Disaster Risk Reduction, chaired by the president, which was called the "platinum level", was responsible for policy decisions. The National Strategic Situation Group (the "gold level"), which included the situation room, analysed, synthesised and interpreted policy decisions. The "silver level" designed planning and operations, and, finally, the "bronze level" was tasked with implementation. The names of these levels mirror military terms for different hierarchical levels of command. Communication between levels and decisions that were taken up and down the ladder caused delays and friction in response efforts. The "silver level" found that decisions sent to the higher levels were often returned without a final answer. At the "bronze level", policies were not always well understood, and responses from the "silver level" were seen as slow, with some decisions being overturned or redirected back to the "bronze level". From outside the state, as an IO representative stated, it seemed they were "working in silos" (IO9 19.01.2018 int.).

Donors' power to negotiate response parameters in a divided state
Aid actors seeking to support the state in the landslide response had to navigate the different, conflicting units of the state. They had to determine with which state institution to collaborate and how to negotiate the terms of the state response.
Sierra Leone's government has always collaborated with international partners. These partners, primarily donors, were able to negotiate a number of conditions in the response. This was seen especially in the prioritisation of cash transfers. As a national state official stated, "That is how they [donors] wanted it because they have contributed a significant portion of that money. And we had no option [other] than giving in to them and do what they want to do" (GOV10 12.01.2018 int.). For one of the donors, the decision to use cash transfers was based on their own commitments and plans to include more cash transfers in emergency programming (IO14 26.01.2018 int.). Discussions and negotiations were conducted at the pillar level between different organisations and state institutions to design a cash transfer programme that would fit into other state-supported schemes. This programme included a mobile money scheme instead of the state-preferred cash scheme, because of donor requirements of accountability and fears of corruption.
In these discussions on cash transfers, the donors used their authority and material resources in a push-and-pull manner, which further divided the state. At the "platinum level", the president initially rejected the mobile money scheme. The scheme was only approved and implemented after a high-level diplomat intervened. Resources were then shared directly with the international implementing partnernot the governmentbecause, according to this donor's assessment, the government did not fulfil the partnership conditions. The international donor was decisive in implementing a mobile approach; however, when a new round of registration necessitated new cash transfers, the donor pulled out, and the state moved to a cash-based, non-mobile approach, motivated by the unused response funds and a more unified approach by the state institutions.
Many aid agencies were supportive of the state's approach, which was carried out by the ONS, but others bypassed state institutions. A national state official reported that the extent to which organisations distributed aid directly presented a challenge to the state's authority: "There were ground rules that no single organisation would go and engage IDPs without going through an official channel. But there were people that were going straight to the IDPs. For us, it actually undermined every effort we are doing towards the response" (GOV10 12.01.2018 int.). Some of the NGOs and civil society organisations especially were less inclined to work closely with state structures. One national NGO set up a small project to support individuals. They contacted these people directly to avoid involving the state, including the chiefs. Similarly, a church hosted IDPs at their school and provided assistance, with limited collaboration with a local state representative. These organisations believed that their power to influence the state was limited, as a civil society representative stated: "For the state, I don't see that room for taking recommendations of churches or NGOs. We don't have much force" (CSO2 10.01.2018 int.).
Especially in post-conflict states, how aid agencies balance their approach with the state is important. Most interviewed national state officials agreed that international assistance was welcomed and that it remained within the parameters of the state's response (GOV8 18.12.2017 int.). However, INGOs referred to the slowness of the state's response and donor requirements as a reason for maintaining control over aid items themselves to ensure efficiency and accountability. The general consensus regarding direct assistance was that the state authorities had to be informed but that distributions could be organised by the organisations themselves, provided this was done in the presence of state representatives.
Not all aid agencies had the same level of influence over the state's response. It was clear that the voices of actors at the central table had more weight than did those of other actors. As an IO representative explained, this was strengthened by both resources and relationships: "Big donors can raise big voices" (IO4 11.05.2017 int.). In a divided and hierarchical state structure, an aid agency's direct relations could create a larger space for negotiating response outcomes. Many organisations collaborated directly with their ministerial counterparts and had little influence in the overall coordination meetings. Others, especially high-level diplomats, had direct influence at the higher state levels. Still, the "big voices" were not always decisive, and strong counter-discourses were also observed. The initial state-donor tensions subsided over the course of the response, as the state was able to control more of the response. As an INGO representative elaborated, "In the end, as in all disasters, people go through a learning phase and eventually the authorities became stronger and took their role. So what was happening was that the authorities increasingly put their feet down and stated exactly what was permissible and what not" (INGO9 16.01.2018 int.).
All interviewees agreed that the state should be in charge of the response. Although the state's capacity was seen as inadequate, the authority and legitimacy to lead remained with the state. In the co-governance of the response, everyone had a role to play, including international partners. However, as one state official remarked, "you do not run an emergency in a democracy" (GOV8 18.12.2017 int.). Some international donors had more power to influence the response, compared with other actors, and this influence was mostly determined by their material resources and ability to work closely with high-level state officials. International partners have a long history in Sierra Leone, which is tied up with postcolonial and post-conflict aid relations and the accompanying power differences.
Community-level response: hybrid governance and the chief-state contention At the local level, some national issues were reproduced, with specific contextual dynamics adding to the complexity. Ceremonial chiefs and community stakeholders acted as mediators of the state at local level. Although they defined their role as part of the local state, they also functioned as societal representatives. This was seen in the contentious relationships between the chiefs and the national state institutions and in the tensions between the chiefs and other stakeholders in some communities. The generational conflict between the traditional authorities and youth became particularly significant in the response, which partly affected the chiefs' role and authority.
Governance at the community level included different state authorities (e.g. representatives of the ONS and the sectoral response pillars) and socio-political authorities representing the youth and women's groups (e.g. chiefs and community stakeholders), as well as other community-based organisations, religious institutions and benefactors. The authority of local structures was recognised by many organisations, which included them in their response. This meant that these local structures functioned at times as a "gatekeeper state", controlling access to aid. However, in the registration, this gatekeeper responsibility was limited by the national state, creating contention between state levels.
The internal competition between different state institutions extended into local governance systems. Chiefs and community stakeholders had started the identification of affected people within their communities before the Ministry of Social Welfare acted. However, when the official registration started, the community lists were not taken into account, and some people were excluded. As an assistant to a chief recounted, "When [the Ministry of] Social Welfare came in, they said they are the eligible institution that is responsible for registration. And no other institution or individual can do registration without them. So they forced us to put aside our registration" (CH3 17.11.2017 int.). The community leaders thus felt bypassed by the state. As outsiders, the state's registration officers were not seen as legitimate or capable authorities to identify affected community members.
The tension between the ONS and the Ministry of Social Welfare was strongly felt at the community level. The authority of both chiefs and state officials is partly derived from controlling resources. Although the ceremonial chiefs consider themselves the local state, their role is not completely formalised and remains unclear. Therefore, both chiefs and state officials had a lot at stake in the landslide response. Controlling the list of eligible beneficiaries, and thus directly or indirectly allocating resources, was important for their political standing. Towards the end of the Ebola crisis, the formal and informal roles of the chiefs were strengthened, but, now, the state's ONS and the Ministry of Social Welfare took full control, undermining the ceremonial chiefs' authority in the community and limiting their gatekeeper role.
Chiefs form part of the internal state at local level as a hybrid institution, but they are often viewed with scepticism. In the interviews, chiefs were accused of partisan behaviour by national state actors: "People were dishonest and the chiefs supported them" (GOV8 18.12.2017 int.). In contrast, organisations that assisted after the landslide often privileged the authority of community stakeholders to identify those affected, as one INGO representative stated, "We don't know the communities; they know themselves" (INGO9 16.01.2018 int.). This sentiment was shared by many, although INGOs were also sceptical about the chiefs' role: "It is an advantage and an added value to have them. But it is also a big risk. Because normally, they have their own interests" (INGO1 16.10.2017 int.). Aid agencies both strengthened and weakened chiefs' gatekeeper role, as these agencies preferred the chiefs' involvement and participation but also waited for the state's official list.
Whereas the chiefs' authority in relation to the state diminished during the landslide response, their traditional authority persisted. To community members, the state's control was an external infringement on their own patrons' authority. However, support for the chief's role varied by community and depended on the chief's relationship with the state. The manner in which the chiefs handled their mediating role affected their role in the response.
One chief's role was considered very positively by community members, neighbouring chiefs, state officials and organisations. In this community, local stakeholders were able to connect with national and international responders and advocate for their needs. Nevertheless, even in this community, the stakeholders felt bypassed in their role as responders. Research participants asserted that the state institutions did not include the stakeholders enough, describing this as though a right had been taken away from them. The stakeholders praised the chief's transparency, which was welcomed by the various state and non-state actors who collaborated with him in the response. However, taking away the responsibility for registration infringed on the role of this chief.
In contrast, in a neighbouring community, the chief's role was contested by the community members, state officials and organisations. Although this chief hosted internally displaced community members at his compound, the role of coordinator of the response was transferred to a community committee, primarily comprising politically engaged youth, which demanded more accountability. The chief was accused of corruption and bias, and he also accused others, blaming the ONS for the takeover by the committee. Although chiefs of neighbouring communities usually collaborate, the neighbouring community chiefs chose not to cooperate with this chief. This chief's inability to control his community members reflected negatively on him, and, with the community committee questioning his authority overall, his authority in the response was also questioned, and his role in the response was limited.
In addition to affecting the chiefs' status, the state response also brought the national state closer to the communities, resulting in friction and resistance. This was mostly seen in the camps, where many IDPs asserted that the selection process was biased and prioritised those who presented themselves first, whereas the ONS insisted that the selection was need-based. IDP participants argued that connections to either government officials or to people associated with the camp increased one's chances of getting in. In Juba camp, one IDP estimated that 80% of the people there were not actually affected by the disaster.
There was a difference in the relations between the ONS management and the IDPs in the two studied camps. Similar to the differences between community views on the two chiefs presented above, the experience of the relationship between state institutions and the people affected how the camp communities viewed the state authorities. The state was seen either as part of the local level or as an outside force infringing on local governance. One camp manager was positively received by the IDPs, but there were tensions with the other camp manager, as one IDP recounted: "When supply comes, she will cease the supply. [. . .] The relationship is not good now because the ONS people, and more especially the leader, has not been nice to us. She has not been treating us with respect" (IDP12 18.11.2017 int.). This resulted in resistance and the non-acceptance of the representation of the state institution at the community level. In this camp, a protest directed at the president was organised in a bid to address community grievances. However, the protest was violently put down by security forces, with one interviewee recounting how she had a miscarriage after experiencing the electric shocks dealt out by the security forces. In their response to this protest, the state was accused of targeting IDPs with whom they had had problems before.
This exemplifies the disconnect between the state, in the form of the ONS, and the IDPs. National state governance at the local level, as seen in the camp management, mostly reinforced a negative perception of the national state, although this differed between the camps. This shows the importance of interactions between local state structures and society. The outcome of the protest reinforced the perceived illegitimacy of the role of the camp management, and therefore also the ONS and even the nation-state by extension. The protesters directed their requests directly towards the president, who had the power to change the realities of the response. This underscores the importance of the understanding of the multiplicity of the state and the complexities of a state-led response.
Conclusion: the local-national conundrum of a state-led disaster response The landslide response affected the legitimacy and institutional capacity of the Sierra Leonean state and its hybrid institutions. The colonial, conflict and Ebola periods have had a major impact on internal state and state-society relations in Sierra Leone. These dynamics were visible in the response to the 2017 landslide. We showed that the policy discourse of state uniformity in disaster response is problematic and that the centralised multilevel governance setting affected state institutions in different ways, with internal state and statesociety politics being negotiated in the response arena.
Governance is not only centred on the state, and questions of power do not concern only state vs. non-state actors. When a policy calls upon "the state" to take up its role in disaster response, different authorities within the state negotiate their legitimacy to do this. The central focus on governments as the responsible entity of the state in disaster response policies is particularly challenging in post-conflict states, where newly formed governance structures and governments are trying to solidify the distribution of power. This invites power struggles not only between the central government and other authorities but also within the national structures themselves.
At the national level, we found that authority between state institutions was negotiated and was a source of contention. Non-state actors had to balance the statecentric support discourse with space for their own interests and voices, raising the question of which state constellation they supported: the unified one in the policies or the divided one seen in practice. The hierarchical nature of governance also created many obstacles in the disaster response, including those related to the relations between the national and local state.
At the community level, each actor's role in the response was negotiated between the chiefs and the state institutions. The chief's authority as a gatekeeper between the state and society is based on both top-down and bottom-up support. In terms of top-down authority, the chiefs were generally challenged because their corruption was stressed in state officials' discourse, but the chiefs' bottom-up authority was largely strengthened by communities' negative perceptions of state restrictions on them. The intermediate form of national state governance found in the camps intensified some state-society tensions, culminating in resistance. Our findings demonstrate that state-society relations are a crucial component of disaster response and that the lines between these two entities are blurred.
These findings are especially important in view of current shifts in international aid paradigms (Hilhorst, 2018). The Sendai Framework accords states the central role in disaster response. In addition, since the Humanitarian Summit of 2016, international aid actors have emphasised the importance of national actors in emergency response, which further underscores the commitment to respect the central role of governments in disaster response. The present study posed several pertinent questions regarding what this means in other post-conflict settings. The themes of intra-state competition and a lack of linkages between different response levels are relevant in all contexts, but especially in situations where state and other institutions are still in flux and the international community has a large role to play. In Sierra Leone, the intra-state competition allowed for a strengthening of institutions after the landslide. However, the disarticulation between the national and local state levels and the state-society tensions have not been adequately addressed.
This article has demonstrated that disaster response affects, and is affected by, a multilayered state with strong internal state power politics. Therefore, current disaster response policy directions are not sufficient. A sensitivity to the internal divisions and competition between state institutions is currently lacking, which may have grave consequences for the internal workings of the state in disaster response.