Humanitarian governance and resilience building: Ethiopia in comparative perspective.

Humanitarian governance is usually understood according to the classic, Dunantist paradigm that accords central importance to international humanitarian agencies. However, this is increasingly paralleled by 'resilience humanitarianism' that focuses, among other things, on including national actors in humanitarian governance. This article views humanitarian governance as emerging through interactions between authorities, implementing agencies and communities. It is based on interactive ethnography in five countries by Partners for Resilience (PfR). Using the Theory of Change (ToC) tool, it analyses the various interpretations and priorities of actors involved in humanitarian problems, solutions and programme governance. For example, PfR had a 'software' focus, aiming to unlock communities' potential for resilience, whereas communities and authorities preferred to receive tangible 'hardware' support. The findings highlight the crucial role of local authorities in shaping humanitarian aid. This is especially pertinent in view of the international agenda to localise aid, which requires the understanding and support of national actors in order to responsibly protect the vulnerable.

Introduction

Humanitarian aid has long been dominated by a classical Dunantist paradigm based on humanitarian ethics. However, in recent years, this paradigm has been paralleled by ‘resilience humanitarianism’ (Ilcan and Rygiel, 2015; Hilhorst, 2018), which links relief to development and focuses on local people and institutions as the first responders to crisis. Resilience humanitarianism first appeared in the domain of disaster risk reduction (DRR) and increasingly underpins refugee care and aid in countries transitioning from conflict to peace. Classical and resilience humanitarianism differ radically as to how interventions are (supposed to be) governed. Governance in classical humanitarianism centres chiefly on the United Nations (UN), international donors, and international non‐governmental organisations (INGOs). Resilience humanitarianism, by contrast, emphasises the role of national and local authorities, service providers from crisis‐affected areas, and affected communities. To further the debate on this new humanitarian‐governance paradigm, this article explores governance in an international programme promoting community resilience.

This Partners for Resilience (PfR) programme was organised by five Netherlands‐based INGOs between 2011 and 2015, and implemented in nine countries. Through a unique research collaboration evolving in the wake of the programme, PfR and a group of academic researchers engaged in interactive research designed and implemented in close dialogue with societal partners. With interactive research, unlike action research, researchers retain their analytical independence. In six countries (Ethiopia, Nicaragua, Guatemala, Indonesia, Kenya, and the Philippines), researchers analysed the programme during three‐ to five‐month periods of qualitative fieldwork, exploring the programme rationale and realities in the different countries. PfR used a model of governance that included the PfR partners, national and local authorities, and the communities where the programme was implemented. Our research focused on how the collaboration between these stakeholders was understood by the different partners, and how it developed over time.

Humanitarian governance can be approached very comprehensively (e.g. Barnett's (2013, p. 380) view of humanitarian governance as pertaining to the ‘global emergence of an order geared to improve people's wellbeing’), comprising an immense web of domains and bodies of practice, extending to development, human rights, peace‐building, and humanitarian aid. Here, however, we focus on an everyday, limited interpretation of humanitarian governance as governing humanitarian action. Although this may appear to be a straightforward sub‐theme, it is not. Humanitarian‐aid budgets are deployed across diverse settings and crises, and there are many differences in humanitarian governance between classical humanitarianism and resilience humanitarianism.

Barnett's (2013) research agenda for humanitarian governance includes four questions that are highly relevant to the present study:

What kind of world is being imagined and produced?

Who governs?

How is humanitarian governance organised and accomplished?

By what authority do humanitarians govern, and what do they do with that authority?

While the first two questions do not specify an actor, the last two put humanitarians at the centre of the analysis. Most humanitarian‐governance literature focuses strongly on international humanitarian agencies, and most studies on humanitarian practice take as a normative and actual point of departure the humanitarian principles and implementation‐styles of large agencies such as the International Committee of the Red Cross, Médecins Sans Frontières, or UN bodies. Other actors, including national authorities and affected communities, appear in the analysis through the eyes of these international humanitarian actors.

Starting from the idea that governance is not simply steered by these international agencies but is also formed in practice by a number of actors including humanitarian‐service providers, governments, and local community figures, Barnett's questions become layered and complex. For example, the first question (‘What kind of world is being imagined and produced?’) can be rephrased to ask what kind of world is imagined by different humanitarian‐governance actors. How are different viewpoints negotiated, and which views gain legitimacy? What contradictions exist between the formal view and views conveyed in everyday reality, and between agreed governance roles and real governance as it unfolds in practice? Rather than assuming that humanitarian action is governed by humanitarian agencies, we advocate studying humanitarian governance as an interplay between humanitarian agencies, national authorities and affected communities (Hilhorst and Jansen, 2010).

Given the layers and variations in governance, case studies are important for analysing humanitarian governance in practice. This article elaborates on the PfR country studies, which used a similar set of questions and analytical framework. We used the Theory of Change (ToC) tool to investigate how different stakeholders framed DRR and how they viewed different actors’ roles in DRR. The findings provide insight into the contradictions, possibilities, and impossibilities of humanitarian governance which includes local actors and communities.

This article is especially relevant in view of humanitarian sector changes heralded during the World Humanitarian Summit (WHS) of 2016. The WHS made a strong plea for localising humanitarian aid, giving greater focus to the role of national governments, service providers, and affected communities in the governance of humanitarian assistance. A major outcome of the WHS was the commitment of approximately 50 key actors to making a ‘Grand Bargain’ in order to accelerate efforts to dispense humanitarian budgets directly to service providers in crisis‐affected areas. Clearly, governance models and analyses of humanitarian action need to become more conscious of these local actors.

We incorporate findings from five of the six case studies conducted.1 To facilitate insight into the dynamics of local humanitarian governance, we present one of these case studies in depth, analysing the other four cases more briefly to identify commonalities and variations in issues of local humanitarian governance. The major case study presented here concerns Ethiopia, where all three authors participated in the fieldwork. The first and third authors coordinated the PfR study, including a joint visit to Ethiopia, and the second author was the primary researcher for the Ethiopia ca se study.

We begin by detailing interactive research, followed by the presentation of PfR's (partly explicit and partly implicit) ToC, which we constructed and validated based on a review of reports and interviews. A section on Ethiopia then analyses how PfR, authorities, and communities each had a different view of the issues and responsibilities in DRR and how this affected the programme. The next section provides brief analyses of the other case studies, leading to the conclusion, in which a number of cross‐cutting issues and concerns are raised.

Interactive research: method and theoretical underpinning

This article originated in a request from the PfR alliance, which was financed by the Dutch government for five years. The partners in this alliance included the Netherlands Red Cross, the Red Cross Red Crescent Climate Centre, Cordaid, Wetlands International, and CARE Netherlands. The programme brought together agencies with humanitarian, climate, DRR, land‐use (ecosystem management) and mixed humanitarian–development profiles. They wanted to develop a comprehensive approach building on their different areas of expertise to enhance community resilience in nine disaster‐prone lower‐income countries.2

This article is based on a study conducted in four phases from 2013 to 2015. The purpose of the overall study was to assess the dynamics and relevance of an integrated approach to building resilience. The research was designed in an interactive way, with the phases of data collection and analysis alternating with phases of reflection with PfR representatives (Argyris and Schön, 1991; van der Haar, Heijmans, and Hilhorst, 2013).

The first phase of the study consisted of a systematic review of the key policy documents of PfR and its country teams. Data were entered and analysed using NVivo data‐analysis software, according to a coding book developed by the researchers and validated by PfR. The findings of this phase were discussed in a workshop that informed the empirical questions guiding the in‐country data collection. After the case studies, which were discussed and validated by in‐country workshops, a general analysis was conducted. This analysis was presented and discussed on various occasions with PfR partners. PfR responded to the final report with a policy brief detailing their feedback on the findings and the policy implications they distilled from the research for incorporation in their next five‐year programme cycle.3

Understanding governance

This article highlights and analyses findings pertaining to humanitarian governance from the six‐country research programme. The concept of governance came into fashion in the 1990s, partly to clarify the layered and fragmented ways state power is used to achieve collective purposes, and partly in response to changing ideas and practices, with collective purposes no longer being solely the domain of the state (Colebatch, 2009). In Europe, Rhodes (1996, p. 57) described the shift from ‘government’ to ‘governance’, where ‘the state becomes a collection of inter‐organisational networks made up of governmental and societal actors with no sovereign actor able to steer or regulate’.

In development processes, the central importance accorded to states in the pursuit of collective purposes had already been eroded since the early 1980s, when non‐governmental organisations (NGOs) came to be considered more effective and intrinsically more inclined to make communities and participatory development the starting point of their endeavours (Edwards and Hulme, 1992, 1996). After a period of acclaim for NGOs (at the expense of the state), it is now more common to build development on a framework of interactive governance, where the formulation, promotion, and achievement of common objectives comes about through different governance arrangements involving multiple types of actor, including the state and NGOs (Torfing et al., 2012, p. 2; Howes et al., 2015; Parker, 2007). Interactive governance has been defined as ‘the complex process through which a plurality of social and political actors with diverging interests interact by means of mobilizing, exchanging, and deploying a range of ideas, rules, and resources’ (Torfing et al., 2012, p. 2).

In industrialised societies, the idea of governance has been used as a frame to understand the changing roles of government; however, in humanitarian aid, governance has rarely been analysed in relation to the state in crisis‐affected countries. ‘Humanitarian governance’ is mainly used to describe how a chain of humanitarian action comes about. Schematically, this chain starts from the international discursive framing of humanitarian crises, moving to the deployment of public resources in donor countries, the implementing machinery consisting of the UN and INGOs, and finally to accountability and participatory approaches at the grassroots level. In critical humanitarian governance studies, the grassroots level is often analysed as instrumental in governmentality attempts, or the shaping of the humanitarian subject according to the image of humanitarian governance (Duffield, 2014).

Only very recently has the notion of humanitarian governance expanded to encompass national authorities and other national institutions shaping humanitarianism praxis. As is often the case in this domain, the discussion has normative, operational, and empirical angles. The normative angle stems from acknowledging the role of national authorities and aid providers in international policy discourse on aid. This discourse is operationalised through interactive governance designs that confer roles to different actors in aid. Interactive governance models are ‘messier’ than models where the government imposes rules. Governance narratives acknowledge the roles of actors inside and outside government, but also point to the blurring of roles and responsibilities, the interdependence of these actors’ actions, and the relative unimportance of command and imposition (Colebatch, 2009, p. 63).

Empirically, we need to look beyond the design of governance to questions of how this works out in practice—what some authors refer to as ‘real’ governance (Titeca and de Herdt, 2011). Different layers of complexity are involved in multi‐actor governance. First, the actors involved in governance arrangements represent different types of institutions, with different normative orders and operating styles, and governance is always negotiated and subject to multiple interpretations (Kraushaar and Lambach, 2009; Putzel and DiJohn, 2012). Studying humanitarian governance requires the development of an ‘antenna’ for these different interpretations and how they stagnate, promote, or change programmes in the course of implementation.

Second, there is a great deal of diversity within the categories that make up governance arrangements. NGOs are often grouped together, but they differ widely in world‐views, mandates, and operating styles. Other examples are local government officers who perform different practices from the central state, or organised community leaders with different views from more marginal community members. Although this could potentially lead to infinite, and therefore impossible, levels of fine‐tuning of our understanding of humanitarian governance, it is important to be aware of these diversities and to bring them into the analysis when they are significant. Finally, governance always has both formal and informal dimensions, and there may be large discrepancies and contradictions between formal and practical norms. Studies of humanitarian governance must therefore address how norms and policies translate into practice (Olivier de Sardan, 2011).

Theories of change as a methodological tool

Our research was informed by the above insights into governance, but we built an analytical framework on applied theories which the partners of PfR were already familiar with. These included the characteristics of a resilient community as identified by Twigg (2009), the Department for International Development's (DFID) livelihood characteristics (DFID, 1999) and the ‘5Cs’ framework (Keijzer et al., 2011). We also used the notion of ToC as a methodological tool to unravel some of the complex layers of governance. ToC, which has been introduced in development programming to replace more conventional log‐frames, maps how an organisation, project, network, or group of stakeholders understand political, social, economic, and/or cultural change to happen, and how they see themselves contributing to that change. ToC aims to define all the building blocks required for the achievement of a long‐term goal. This includes the smaller outcomes needed for larger change, as well as the interventions thought to lead to these outcomes. Importantly, a ToC also clarifies how interventions and outcomes are understood to be linked (Anderson, 2006).

To make our conversations with PfR actors about the ToC useful in our exploration of governance, we introduced two changes to the managerial notion of ToC. First, we introduced the idea of ‘theories of change’ to emphasise that the authoritative or formal ToC is subject to interpretation by different actors and may not be considered legitimate by all actors. Second, we emphasised that, although a ToC represents the story an entity tells about its context, its contribution, and itself, this may not correspond to the ToC conveyed through the organisation's practices. In terms of governance theory, ToC represents formal norms and practices; it is necessary to establish empirically how these relate to practical norms and practices.

Research question and fieldwork methodology

This article addresses the question of how humanitarian governance, understood as multi‐actor arrangements to deliver humanitarian services and goods, comes about when resilience is promoted in local contexts. To make the analysis possible, we focused this study of governance on interactions between humanitarians, authorities, and communities.

Knowing that different actors may work through different logics and have different ways of framing the ‘problem of DRR and lack of resilience’, we asked how different actors responded to PfR's ToC. What did they identify as the main humanitarian problem? What objective did they want to attain? How, by whom, when, and where could this be achieved? In the initial phase of the PfR programme, we observed that PfR actors tended to distinguish sharply between so‐called hardware approaches (i.e. predominantly transfer of materials and tangible field‐activities) and software approaches (i.e. working on social structures and processes, and skill transfers). We maintained this distinction in the interviews. In addition to collecting different views on how the programme should be carried out, we also asked how the programme worked out in practice.

In each of the country studies, national‐level actors were first interviewed—individually, through focus‐groups and in workshops. In dialogue with the research participants, the researchers then identified multiple case studies to investigate the alliance in action at the local level, where government officials, community members, and NGO staff members running the field operations were approached. The subcountry cases varied to some extent: they usually comprised geographical areas, but there were also several specific activities and campaigns selected for investigation.

Data gathering

Data gathering in the five countries largely followed the same pattern. Here, we elaborate solely the process for Ethiopia, where the four case‐study sites were widely scattered across the country. As Figure 1 shows, the case study sites were selected to represent different implementing partners, agro‐ecological zone types (the Borena, Afar, and Dire Dawa lowlands, with different forms of agro‐pastoralism, and the farmed highlands around Ibnat in the Amhara region), timespans of PfR implementation (inception in 2009, 2011, and 2013), and activities.

Figure 1

Map of PfR implementation sites and case‐study sites in Ethiopia

Source: modified from Desportes (2015, p. 20).

In all these sites, field staff, community members, and government officials from the kebeles (lowest‐level governance) and woredas (district governance‐level) were asked about their framing of the problem, interpretation of their role and the roles of others, and perceptions of the programme. In‐country NGO staff members facilitated the fieldwork logistics, but, as much as possible, the interviews were one‐on‐one, without the presence of NGO or government observers. The federal government's perspective could not be directly included in 2014, but we also draw upon an interview with a federal government official conducted by the second author in 2017 as part of another humanitarian governance research project.

Secondary sources such as project‐monitoring reports, local risk assessments, action plans, and environmental impact‐assessments were also analysed to gain a full understanding of the programme rationales and the practicalities of its implementation.

PfR's ToC

PfR did not have a ready‐made ToC, but it did have many policy ideas about how the programme should work, with building blocks, working principles, and objectives. Based on a content‐analysis of reports and a workshop with the Netherlands‐based alliance coordinators, we constructed and validated the following ToC as the principal storyline of the alliance:

1) Communities face major and multiple disaster risks.

2) Reducing these risks requires strengthening community resilience, and therefore communities should be central in the programme.

3) To reduce people's vulnerability to the impact of hazards, it is important that they can anticipate, respond to, adapt to, and transform disaster risk.

4) Enhancing livelihood opportunities is important and should be part of risk‐reduction activities.

5) To reduce the root causes of disaster risk, ecosystem management and restoration (EMR), and climate change adaptation (CCA) need to be integrated into DRR.

6) A successful integration of DRR, CCA, and EMR requires working across different geographical areas and timescales.

7) The integration of different approaches to working towards community resilience requires strengthening the collaboration between multiple stakeholders across sectors.

8) Because ‘disasters’ usually result from processes beyond the locality of manifestation, they require solutions that are not in the hands of communities alone, so it is important to involve stakeholders at different societal levels, including the government and research institutions.

9) Because of the complexity of the problem and the programme there is a strong need to be adaptive.

This ToC shows that multi‐actor governance is a central element of PfR's approach. Starting from the understanding that supporting the resilience of communities requires a comprehensive approach, the alliance wants to strengthen collaboration between multiple stakeholders across sectors. The alliance also emphasises that this multi‐stakeholder collaboration must happen at different societal levels. This ToC was constructed and validated with the PfR steering‐groups before the start of the fieldwork, and hence could be used as input for the country studies.

Findings from Ethiopia

In Ethiopia, PfR was carried out from 2011 to 2015 by an alliance comprising the Ethiopian branches of Cordaid and CARE, the Ethiopian Red Cross Society, and five local partners4 (hereafter referred to as ‘implementing partners’, in contrast to the ‘alliance partners’). Wetlands International and the Red Cross Red Crescent Climate Centre provided technical advice and training from their regional centres. In addition to policy and advocacy activities to disseminate lessons learned and facilitate project take‐over, the PfR partners carried out various activities in their nine project implementation sites, as summarised in Table 2.

Table 2

PfR Ethiopia categories of interventions

Intervention	Examples
1. Ecosystem management and restoration	Rangeland rehabilitation through enclosures, hillside rehabilitation through terracing, soil and water conservation with soil band and check dams
2. Livelihood diversification	Setting up savings and loan associations; cooperatives producing aloe vera soap, honey or energy‐saving stoves; distributing goats or chickens; providing irrigation infrastructure such as ponds, dams, canals, or drip irrigation; distributing seed and farming tools
3. Establishing and strengthening community institutions	Committees responsible for managing disaster risk in their kebele, with some also acting as local early‐warning committees Committees managing specific PfR interventions (e.g. an irrigation committee) Environmental school clubs
4. Training	Theoretical and practical skill transfer, for instance concerning farming, meteorological forecasts, and accounting (including specific capacity training for the newly established community institutions) Training in the form of exchanges between (implementing) partners, government officials, and communities, with joint risk assessment and development of action plans

Source: Desportes, 2015, p. 24.

Before detailing how community and government members aligned or did not align with PfR's ToC, it is useful to describe the broader country context. As far as relations between the government, communities, and NGOs in Ethiopia are concerned, development and humanitarian endeavours are largely government‐led. This has been the case since the implementation of large‐scale environmental restoration programmes in the 1980s, where community members were ‘put to work’ restoring their ‘overused’ and ‘degraded’ environment, following the neo‐Malthusian narrative deconstructed by Hoben (1996). This large‐scale government‐engineered development approach conforms to the high‐modernist approach of Ethiopian statebuilding, which strongly asserts state sovereignty and ‘control over natural resources and the people that use them’ (Fantini and Puddu, 2016, pp. 97–98). State control is also asserted over continually diminishing numbers of civil‐society organisations and NGOs, largely viewed by the government as a parallel and inefficient resource‐provision system (Vaughan and Tronvoll, 2003). NGOs’ ‘recurring difficulties in negotiating presence, access, and operational space’ (del Valle and Healy, 2013, p. 189) were exacerbated following the contested 2005 elections and the added administrative restrictions of the 2009 Charities and Societies proclamation (Carruth, 2016; Corbet et al., 2017; International Centre for Non‐Profit Law, 2012). However, to some extent, international donors and NGOs have adjusted to government control. As Hoben (1996, p. 197) noted, ‘the humanitarian, community, and environmental emphases’ of large‐scale modernist environmental programmes made them ‘live with, and even appreciate, the top‐down and even authoritarian way’ in which the government controls them.

Diverging ToCs

Table 3 summarises the major ToC building blocks (problem definition, description of the goal, and specifics of ‘how’, ‘who’, ‘when’, and ‘where’ to attain that goal) by three major kinds of actor types (PfR staff, government, and communities), although internal variations did of course occur.

Table 3

Major ToC building blocks by actor type

ToC building block	PfR staff	Government	Communities
Problem statement	Hazards Degraded ecosystem and various pressures on it Lack of agency (aid dependency)	Hazards Environmental degradation Lack of financial resources Communities’ poor mindsets	Lack of resources and opportunities Hazards and rain dependency
Goal	• Resilience (i.e. livelihoods protected and growing sustainably)	• Decreased need for emergency aid, stable livelihoods	• Livelihoods protected and growing sustainably
Solutions: What/how	Empowerment predominantly via software, but also hardware Creating an enabling environment (advocacy, commitment, structures, plans) Environmental restoration	• Hardware especially (no excessive meddling in local structures)	Hardware support Increased practical knowledge and skills
Solutions: Who	Community‐driven Government‐supported	Communities’ labour (not necessarily ideas) NGOs (following government approach) Government	NGOs Government
Solutions: When/Where	Short‐term Wider landscape Engaging supra‐local governance levels	Short‐ and long‐term (not bound to political term) Locally centred	Short‐term Community‐centred

Source: modified from Hilhorst et al. (2015).

Problem definition

Table 3 highlights some similarities but also many differences in ToC perspectives. Although all actors mentioned the strenuous environment and recurrent droughts as major problems, PfR staff members emphasised a combination of environmental factors and attitude‐problems at sites with a long history of NGO presence. Government officials stressed the problem of dependency, whereas community members spoke mainly of acute poverty and difficulty surviving in a degraded environment.

Goal definition

Definition of the overall PfR programme goal varied among staff members, with a more basic form given by field staff (e.g. reducing poverty or reducing the impact of natural hazards on livelihoods) and more holistic representations from the Addis Ababa‐based office staff. In its most detailed form, the desired goal is a resilient community characterised by ‘embedded’ and sustainable livelihoods. Embeddedness echoes the EMR component. As one staff member explained, it means that ‘livelihoods must be grounded in people's environment in such a way that the surrounding resources are sufficient to support them’.5 Sustainability was viewed mainly in relation to CCA, with viable livelihoods in the long term, resisting climatic variability. DRR, CCA, and EMR are thus integrated in PfR's ToC and linked to livelihoods.

Community understanding was often similar to that of PfR staff, as reflected in this statement from a male community member in Ibnat (1 October 2014):

The aim is to alleviate the risks which acted on us previously, such as soil erosion and flood and drought. The project enables [us] to retain our wealth, our livelihood.

Although government officials also aimed for stable livelihoods, they stressed that this would lessen the need for emergency relief. Their understanding was far less integrated than the view of PfR. They tended to conceptualise change in sectoral terms, such as objectives for agriculture or irrigation, matching the government's sectoral organisation.

Solutions: how

In explaining how resilience could be achieved, Ethiopian PfR staff members mainly emphasised the roles of the community. A community first needed to be empowered through participatory risk assessments and action plans, as well as by software activities such as training and the formation of community committees. An empowered community was presented as aware of its environment and how it could act within it. In their focus on communities, alliance partners emphasised the development of social institutions over providing material assistance. Several staff interviews made it clear that their agencies believed institutional development was key for food security because it would unlock resources available in the community—notably local knowledge. Staff members from agencies with a more humanitarian background departed slightly from this dominant understanding, giving more space to hardware support for community activities.

Whereas the PfR partners emphasised the strengthening of community institutions as a precondition for resilience, the government and communities stressed the importance of the provision of material support. This difference was apparent in interviews, leading to discussion on programme implementation. For example, a woreda government official from a Borena cooperative office complained that PfR did not release funds for a qualified external accountant for the savings group, as is done by other NGOs in the area.6 PfR wanted community members to learn this activity themselves in order to be independent in the long term.7

There were also commonalities in the ToCs. In the PfR approach, empowered communities to regain and fortify a natural asset base via community‐based natural soil and water conservation activities. This clearly resonated with the federal government's strong focus on environmental resource management in recent decades.8 Communities’ appreciation for this environmental approach differed depending on how long they had been involved, increasing over time. A community which had worked with this approach since 2009 had seen more tangible results, including in terms of livelihoods, than had one where PfR began more recently.

Although there was significant agreement about the steps required, ideas on how these should translate into practice differed radically. All parties, for example, agreed that livelihoods should be diversified to improve resistance to shocks and stresses. In the case of pastoralist communities, however, government policies tended to be geared towards (forced) resettlement into consolidated centres, which could become centres of livelihood activities. These communities were mostly opposed to this idea, instead favouring alternative options supporting livestock‐rearing and small‐scale diversification. PfR staff members emphasised the need to respect communities’ choices and traditional ways of living. The PfR implementing partners were more diverse in their attitudes towards pastoralism, with some leaning towards government policies and others towards the communities’ perspective, as well as engaging more openly with parallel pastoralist traditional governance systems. In practice, the programme took on different shapes in different areas, depending mainly on the local implementing partner and their relationship with the government. Some local implementing partners could in fact be classified as ‘government‐oriented NGOs’, acting as a local arm of the government.

Solutions: who

Ideas about who should play the lead role on the path towards resilience aligned closely with the actors’ preferred path. According to Ethiopian PfR staff, empowered communities should take the driving seat in building resilience. This was considered effective (grounded in local context), efficient (involving locals from the start increases community acceptance), sustainable (built on local structures), and cost‐effective.

Moreover, PfR aimed to move away from competition among agencies ‘working in silos’ to reach their objectives towards partnership and learning. On a wall in a Borena DRR community group office set up by PfR, a poster (Figure 2) unpacked the ‘big shift’ deemed necessary to implement the programme. Instead of dominating and teaching, programme staff should facilitate. They should not extract but empower, preferably with visual means rather than by sending messages. Relations should be open, and people should think about the community.

Figure 2

‘Big shift’ within which PfR is embedded

Source: Desportes 2015, p. 42.

A striking difference in the ToCs is how government officials did not see communities as key in the bottom‐up identification of needs, but rather as the lowest level of a top‐down governance model and a workforce to implement activities. Interviews with local woreda and kebele government officials indicated that, rather than listening to the communities’ views, these officials saw working with the communities as an opportunity to teach them how to implement (government) development programmes. Several government officials in Dewe and Ibnat highlighted that the role of community committees was not to take decisions, but to ‘provide information’.

Having been subject to top‐down governance structures for decades, community members had internalised the government perspective, looking towards governments and NGOs to provide necessary material support.

Solutions: when and where

Whereas local government actors and community members saw the programme as a local, time‐bound endeavour, PfR members wanted the resilience‐building experiences to extend beyond the end of the programme and to spread to other areas to achieve longer‐term, larger‐scale change. This is why advocacy was one of the alliance's strategic objectives.

Still, we observed variation in the entry‐point to advocacy among PfR staff members. The Netherlands‐based PfR partners often explained advocacy using rights‐based jargon, but Ethiopian partners were very careful not to frame advocacy in political terms. PfR Ethiopia maintained an apolitical profile by refraining from advocacy beyond bringing practical community needs to the government's attention. Community members such as the pastoral networks were likewise trained to propose their own solutions and advocate for their needs with local authorities. The Ethiopian Red Cross Society was especially cautious not to jeopardise its auxiliary position with the government. One staff member explained (22 September 2014) its approach as follows:

The government is fine with us establishing committees. Most did not know this approach before. When we teach about it, they are happy, appreciate the philosophy behind it. It fails only if it is associated with politics. We need to focus on practical problem‐solving […] It is invisible politics.

Theory versus practice

Despite PfR's strong discourse on communities being in the driving seat, the poster shown in Figure 2 hints at a different reality. The handwritten poster reflects the isolated and impoverished environment in Ethiopia's southern periphery. However, in a community where people speak Oromifaa, the poster was written in English. This raises the question of how its message might be translated into practice, and also illustrates the governance relations in the programme.

On the one hand, the programme employed a participatory approach, placing a premium on decision‐making in the community. The following attitude expressed by an Ibnat resident (1 October 2014) was not unusual:

We don’t believe the Red Cross will continue with us. It will leave us. But it gave us the capacity, the knowledge, technical, materials, in order to handle things. We acknowledge it with good mood. We have a committee in place, a water management system. We did the activities without payment until now. We can continue.

On the other hand, communities might have different priorities from those facilitated by PfR. The communities often expressed a preference for hardware support, but PfR seemed to actually occupy the driving seat, working to make communities ‘ready’ to internalise PfR's emphasis on soft interventions geared towards institutions and training.

Over the course of the programme, it also became apparent that government buy‐in was lower than PfR assumed. Although the government shared PfR's conviction that communities were important, they saw communities as the end‐line of sectoral development programmes, and NGOs as resource‐providers and facilitators for those programmes. A federal government official reflected on community‐based early‐warning systems (9 June 2017), which were also implemented as part of the PfR programme (with efforts to link early‐warning committees with government offices):

[NGOs] can feed us [with ideas], but they should use the common platforms. For instance, I would love to have one strong early‐warning system in the country. International NGOs get resources to do their own community‐based early‐warning systems, etc. Funders love that. But I tell them not to do that, to go with us. So we can create one strong system. They invite me to workshops, but I do not go.

Although kebele and woreda government officials tolerated the community committees, this did not change the government's approach to resilience. The committees remained isolated, with no official legal basis, and unable to advocate for government services. This hampered the scaling‐up and sustainability of the approach. Although the committees and the participatory style of PfR were appreciated, tangible outcomes were lacking; the knowledge and value of the PfR approach did not reach the regional government, so no non‐NGO funds or efforts supported the new structures or follow‐up on activities. A former PfR staff member who visited the Borena PfR site in 2017 shared the unsurprising news that the local committees had stopped operating after the programme ended, as did their facilitation and support of the communities.

Findings on resilience building: comparative case studies

In this section, we compare the findings on Ethiopia with findings from four other case studies. The comparison focuses on the PfR objectives to (i) unlock community potential, and (ii) advance the capacity of community institutions to advocate for government or NGO support to fulfil needs that were beyond community resources. We also bring into the comparison PfR's focus on software components, such as the formation of local planning committees and awareness‐raising activities, to increase community ownership of the programme. Although the set‐up of the programme across countries was similar, there were differences in how it played out in the governance constellation of implementing agencies, government, and communities, and in the challenges encountered.

Nicaragua (based on Strauch, 2015)

There were many similarities between PfR Nicaragua and PfR Ethiopia. Nicaragua has a very different history, with civil society strongly engaging with and politicising processes of DRR (Morales Carbonell and van der Haar, 2017). Nonetheless, where PfR operated in Nicaragua, the space for civil society to express views and engage in political debate was shrinking. This may have been specific to these geographical areas, as they had long been affected by drought and had an extended aid history. The government displayed a paternalistic approach towards communities, and state actors and PfR staff were convinced that the population suffered from ‘dependency syndrome’. Both the government and PfR considered communities’ passive attitude the major obstacle to resilience. This strengthened the emphasis on software approaches for behavioural change.

The programme focused on identifying good CCA and DRR practices based on indigenous knowledge and strengthening these with scientific knowledge. PfR activities had many positive impacts in communities; however, as in Ethiopia, the potential for achieving structural, transformative change was uncertain because community initiatives were rarely scaled up. This could be attributed to fragility and a lack of autonomy in local Nicaraguan committees, which were government‐imposed structures affiliated with the governing party. Possibilities for a bottom‐up policy and advocacy approach were limited, as political instability and a fragile social fabric added to uncertainty and restricted space for policy‐implementation.

Guatemala (based on Dávila Bustamente, 2015)

PfR Guatemala was implemented in five risk‐prone departments which have had high aid density since the end of the civil war in 1996, usually in the form of humanitarian handout programmes. Perhaps this was why communities were initially highly averse to PfR's software components. Some communities demanded material support from PfR and refused to participate in a project without such a component. Paradoxically, by refusing to adopt the PfR approach that put community institutions in the driving seat in the programme, the Guatemalan communities steered the programme more compared with the Ethiopian communities which followed the PfR model. The refusal to participate in participatory development can be a strong token of self‐determination (see Cooke and Kothari, 2001).

An in‐country survey (n = 145) indicated that residents valued the impact of material support to households and communities most, followed by improved social ties, environmental restoration and knowledge about integrated risk management. Among the different project approaches of implementing partners, ‘bio‐rights projects’ stood out in terms of community appreciation. In these initiatives, community groups received micro‐credit loans and training tailored to projects they devised themselves. The groups repaid their loans by engaging in ecosystem‐based risk‐reduction measures. The projects thus combined community empowerment, material transfers, and environmental and livelihood improvements.

Guatemala was the only case where policy and advocacy efforts were fruitful at national level. The PfR programme resulted in a strategic inter‐institutional agenda to align the plans and actions of key government actors. The NGOs’ previous advocacy experience in the country made the success of this programme‐component possible. Very early in the programme, the partners developed a roadmap for national‐level advocacy.

Indonesia (based on Srikandini, 2015)

In Indonesia, PfR had a different trajectory. It was implemented during major national DRR reform: Indonesia doubled its national DRR budget and created district‐level multi‐stakeholder DRR forums. The conditions thus seemed very favourable for PfR to influence policy and practice towards a comprehensive DRR approach grounded in environmental management, CCA, and development. However, this potential was not realised because the PfR partners were never able—nor did they even aim—to align their approaches. The Indonesian PfR partners lacked the Guatemalan partners’ experience in planning and performing advocacy, and preferred to continue their community programmes as usual. The lack of alignment among PfR partners was partly related to the country's size and its multitude of islands, and was not dealt with through monthly meetings or exchange visits, as was the case in Ethiopia.

The programme was also not integrated at implementation level, but projects in the 37 Indonesian PfR implementation villages usually addressed one or more components of PfR, depending on the expertise of the implementing partner. Implementers seemed to treat PfR primarily as a funding opportunity to implement their own programmes. However, during the PfR implementation, a common emphasis on local‐level advocacy and training emerged, where community members approached the local government (facilitated by PfR) to present their DRR‐related needs. However, these local advocacy skills did not result in increased responsiveness by the local government, and the increased national‐level DRR budget did not trickle down to the local level. As in Ethiopia, local government officials who, influenced by PfR, became sympathetic to local DRR needs felt incapable of meeting these needs because they had no budget for DRR.

Kenya (based on Faling, 2015)

PfR Kenya was implemented in 13 (semi‐)arid areas along the Ewaso Nyiro River in Isiolo County. As in Ethiopia, PfR initially focused on community‐level capacity‐development rather than material assistance. Instead of forming new committees, the programme relied on existing community development committees established under a previous World Bank programme implemented to streamline and coordinate different aid initiatives in the region. Frequent exchange visits took place between the communities.

A Kenyan field staff member described the project as follows:

PfR is only a project. But the way it is implemented is very sustainable. The seed has been planted and the communities are appreciating the approach. Initiatives are increasingly coming from community members themselves.

The programme coincided with extensive national policy reforms aiming for decentralisation and improved coordination among the various bodies involved in DRR. As in Indonesia, PfR used this as an entry‐point to advocate for community‐based integrated risk management. Conditions in Kenya were more favourable for a networked type of governance, but this also led to particular challenges: working through existing community‐organisations reinforced existing power‐relations and reduced PfR's leverage in the implementation‐process.

Beyond the community, an umbrella organisation linking several civil‐society organisations in Kenya was an effective medium for advocacy. The network organised a 250‐kilometre camel march to challenge the national and county governments on the negative environmental and social impacts of a mega‐dam that was planned upstream. The civil‐society organisations also managed to engage private companies, the local and national media, and some elements of the government in supporting the Ewaso Ngiro River communities’ cause. This would have been unthinkable in the authoritarian Ethiopian governance context a few hundred kilometres to the north. Humanitarian governance, in the Kenyan case, developed into a strong alliance between PfR and communities in their effort to mobilise the state.

Discussion

Humanitarian governance is often treated as a single reality, seen from the point of view of international humanitarian actors. This article has viewed governance as emerging through interaction among international and national, state, and non‐state actors. We have analysed governance primarily as the interplay between the state, implementing agencies, and communities.

More specifically, we explored humanitarian governance in the scenario of resilience‐building programmes. As the cases demonstrate, this type of scenario is far removed from the high‐conflict aid scenario that usually informs discussions on humanitarian governance. Nonetheless, in view of the turn to DRR and the promotion of resilience in addressing natural disasters (see, for example, the Hyogo and Sendai Frameworks for Action of 2005 and 2015, respectively (United Nations International Strategy for Disaster Reduction (UNISDR), 2005, 2015), this scenario represents a widespread and substantial body of practice in the humanitarian domain (Kelman, 2017).

It is important to recognise that the multiple actors involved in governance have different understandings of the aims and rules of operation. In this research, we developed a tool based on theories of change as an entry‐point for our enquiry. ToC language enhanced the interactive nature of the research and enabled dialogue with PfR stakeholders who were already familiar with the ToC jargon. Although governance issues varied greatly across the countries, a number of striking patterns appeared.

Grounded in five in‐depth country case studies of a large programme to enhance resilience and DRR (Hilhorst et al., 2015), this article revealed one major finding. Actors in the PfR agencies, local authorities, and communities often had different interpretations of the ToC and different priorities concerning its implementation. PfR activities tended to have a strong ‘software’ focus, aiming to unlock communities’ potential to become more resilient, but communities—and often local authorities—preferred to receive more immediate, local and tangible ‘hardware’ support. This difference was especially pronounced at the beginning of the programme. One way of framing this issue is to see the aid organisations in the PfR alliance as having changed their approach to resilience humanitarianism, whereas the local actors preferred more classic forms of handout humanitarianism (Hilhorst, 2018). Clarifying and bridging differences in objectives and expectations are thus pivotal parts of the humanitarian‐governance agenda.

The findings paint a mixed picture of the power and normative rationale of intervening agencies to shape programming. In the context of intervention, a programme like PfR may be experienced as imposing its ideology, even when that ideology places communities’ views at the centre. In several countries, community members expressed a different preference for addressing their resilience, and implementing and national partners felt that the programme was organised in a top‐down fashion by the national or international headquarters, respectively. However, PfR's power to shape the programme to achieve its objectives seemed limited in many ways. PfR seemed a coherent programme on paper, but, in the complexity of the intervention, there was an abundance of different attitudes to the approach. The Dutch agencies, their in‐country national counterparts, the authorities, and communities each brought their own ideas and interests to the programme implementation. Unsurprisingly, PfR struggled to realise its mandate amid the messiness of multi‐actor humanitarian governance, where the programme was subject to local realities, interests, and collaborations.

The findings highlight the importance of analysing the different ways governing actors view the programme. Particularly striking are the findings about governments. Programmes like PfR tend to allot a critical role to the government as a key actor which could turn DRR into a sustainable, structured, and scaled‐up endeavour, if government actors can be convinced to ‘see the light’. Across the countries studied, we found many factors cautioning against an optimistic take on this. Despite apparent buy‐in on a ToC giving central importance to communities, in practice the Ethiopian government had a very different understanding compared with that of the PfR partners. These differences ultimately prevented the scaling‐up of an integrated approach to DRR. For various reasons, including high turnover in local government in the Philippines, a local Indonesian government that had hardly any budget and was therefore ‘sitting on a dry desk’ (Therkildsen and Tidemand, 2007, p. 8), and a lack of coordinated planning in Nicaragua, it was rare that individual government actors ‘seeing the light’ would result in changing actual governance practice.

PfR's experience provides several ideas on how programmes aiming to realise resilience can deal with these situations. One chosen strategy was to seek convergence with other policy agendas or national thematic priorities, such as ongoing policy reforms. Another strategy was to carefully explore the room for manoeuvre to position the programme within existing governance traditions, navigating the prevailing realities to achieve some of PfR's goals. Finally, the duration of the programme could be extended, so that different actors might implement, internalise or adapt the hopedfor changes over a longer period of time.

The findings remind us that, in humanitarian programming, a localisation agenda requires investing in understanding, negotiating, and supporting national actors to responsibly implement humanitarian programming for vulnerable people. When we reformulate the final conclusion from the point of view of the governments in the research, the findings remind us that intervening agencies should avoid setting unrealistic objectives that would require great changes in existing priorities and historically grown styles of governance. Reformulating the same conclusion from the point of view of the communities studied, we are reminded that outsiders intervening on behalf of communities should start listening to those communities if they want to achieve sustainable results.

Acknowledgments

We sincerely thank the Partners for Resilience (PfR) for their openness in this research, with special thanks to our PfR liaison Raimond Duijsens. We also acknowledge the contribution of our fellow researchers on PfR: Marijn Faling, Lisa Strauch, Yvonne Leung, Mariana Dávila Bustamente and Annisa Srikandini. This article was made possible through a Vici grant from the Netherlands Organisation for Scientific Research (NWO), grant number 453‐14‐013.

Table 1

Overview of respondents and data collection methods, Ethiopia

Method and respondent category	Number
Semi‐structured interviews with alliance partners	8
Semi‐structured interviews with implementing partners	7
Informal discussions with community members, most often at project sites	47
Focus‐group discussions (n = 16) with community members in seven kebeles	129
Semi‐structured interviews with kebele government officials	9
Semi‐structured interviews with woreda government officials	18
Semi‐structured interviews with regional government officials	1
Semi‐structured interviews with external stakeholders (e.g. university staff, in‐country consultants and NGO facilitators)	6
Observation at the Ethiopian Red Cross Society's ‘Enhancing system and community resilience’ panel discussion (16/08/2014), key stakeholders (e.g. UN official)	5
Total	230

Source: modified from Desportes (2015, p. 26).