Reproducing remoteness? States, internationals and the co-constitution of aid ‘bunkerization’ in the East African periphery

ABSTRACT The physical and social retreat of international interveners behind the walls of ‘bunkered’ aid compounds in (putatively) more remote and dangerous regions of the South has been the focus of growing critical attention in recent years. An increasingly remote and fearful culture of risk aversion and differentiation among Western states and organizations has been largely identified as the driving force behind this set of practices. This article presents a different perspective on the bunkerization phenomenon through focusing on the agency of Southern states in the process. Exploring bunkerization across eastern/central Africa—and in Ethiopia’s eastern Somali region in particular—the study emphasizes not only how African states have been key promoters of modern bunkerization, but also how bunkerization behaviour and mentalities have historically characterized how many African borderlands—and contemporary sites of international intervention—have been incorporated into the global state system.


Introduction
No image better captures the apparent physical and emotional withdrawal of the international aid and peacebuilding industries from the communities they seek to assist than that of the 'fortified aid compound'. Perhaps most comprehensively and powerfully unpacked by Mark Duffield in the pages of this journal (Duffield 2010), the militarized structures which house and 'protect' UN, International Non-governmental Organization and other international interveners in South Sudan, Afghanistan, Somalia and elsewhere stand as a commanding metaphor for a conscious and fearful demarcation and defence of a separate 'lifeworld' (Rossi 2006) by 'internationals' in the midst of a foreign land. These compounds contain and secure the comforts and accoutrements of everyday life for their inhabitants (Higate and Henry 2010;Smirl 2015). Access is, however, restricted to a pre-cleared class of international associates-those who dwell within and subscribe to Severine Autesserre's 'dominant peacebuilding culture ' (2010, 2014)-with local actors not simply excluded but posited as unpredictable and threatening.
For these and other scholars, the compound is the manifestation of a much broader process of mental and physical 'bunkerization' undergone throughout the international aid community in the context of an increasing militarization and securitization of development since the 1990s. A growing loss of 'ground truth' among interveners has followed, it is suggested, founded ultimately in risk aversion and a decline in solidarity between the North and South (Duffield 2010(Duffield , 2014Sandstrom 2014). Crucially, it is argued, this process is a relatively new one (Duffield 2014, S82-S83).
The purpose of this article is to propose an alternative lens through which to understand international remoteness and aid militarization, critically examining the role of Southern states themselves in the aid bunkerization phenomenon. In doing so, the article unpacks the linkages between historical patterns and practices of statebuilding on the one hand and contemporary international bunkerization on the other, focusing particularly on central and eastern Africa.
For in much of eastern Africa, state expansion, consolidation and governance by colonial, post-colonial and proto-colonial polities since the nineteenth century has contained marked similarities to those patterns of behaviour highlighted above. Indeed, governance through militarized encounters, garrisoned outposts and chains of local 'brokers' is not purely a feature of the contemporary aid industry in the peripheries of Sudan, Uganda and Ethiopia but, in many respects, how the state and its technologies themselves came to parts of the African borderland.
The drawing of this link is not intended as a device for suggesting that today's interventionary practices in the South are simply an unconscious extension of underlying, historical practices of-often exploitative-governance. Instead, the aim is to feel a new path in the current literature on bunkerization which has, to date, largely overlooked the role of Southern states in driving the phenomenon and in shaping Western encounters with, and thus understandings of, many Southern communities. This is undertaken in the second part of the article through an exploration of contemporary 'bunkerized' interventionary practices in the Somali region of eastern Ethiopia (Somali Regional State or SRS). This latter section of the article draws upon interviews undertaken with senior, midlevel and operational officials in UN agencies, Western embassies and aid agencies, INGOs, implementing bodies, consultancy groups, risk analysis outfits and Ethiopian state officials between March and May 2014-all of whose focus has been on activities in SRS. Almost all respondents requested anonymity owing to the perceived political or commercial sensitivity of the information being provided (on security protocols, risk assessment procedures, development of funding bids etc.). Consequently, while as much information as possible is provided, interviewees are cited without reference to names or, in some cases, organizations. Where the same designation is used for multiple interviewees, separate interviews are indicated with numbers (e.g. 'Interview with UN official 2'). The remainder of the paper builds upon secondary literature along with interviews with international and state officials undertaken across the region in 12 fieldwork visits since 2009.
In terms of case selection, the choice of the central/eastern African region reflects the heavy concentration of international intervention and bunkerization behaviour there since the 1990s. The comparatively high number of attacks on, and killings of, aid workers across this region-conceived of, by many, as a key driver of bunkerization protocols and bunkerized living by internationals-is also a central consideration. 1 The varied nature and strength of the pre-/post-colonial and colonial state across the region also provides a valuable differentiation of contexts in which to compare and contrast the relationship between statebuilding, intervention and state 'hardness'/'softness'. For while the Ethiopian case study alone might be seen as exceptional, given the historical strength and territorial reach of that state's political and security machinery, the fact that similar phenomena can be identified elsewhere in the region in states with historically much more distant relationships with peripheries underlines the broader regional generalizability of the article's findings.
The article begins by unpacking the literature on bunkerization before drawing parallels between practices identified here and those undertaken in many east African border regions during the nineteenth and twentieth centuries. The SRS case is then analysed with a particular focus placed on the role of the contemporary Ethiopian state in promoting bunkerization practices among modern international interveners. The piece concludes by arguing for a greater appreciation of the role of states-and historical practices of statebuilding-in scholarly analyses of international intervention and knowledge construction.

A note on terminology
The terminological characterization of different categories of actors at intervention sites stands at something of an impasse in contemporary critical peacebuilding scholarship. The binary distinction drawn by many policymakers and scholars between 'international' and 'national'/'local' is problematic in its degree of imprecision and essentialization (Heathershaw 2013). The lines between staff of donor institutions, aid agencies, INGOs, contractors and other 'aid workers', for example, are exceptionally blurred in many intervention sites, not only because the latter are often the implementers of the formers' programmes but also because 'local' or 'national' staff frequently play a prominent role in many donor/INGO institutions in these contexts (Collinson and Duffield 2013).
Likewise, the notion of a 'local' or 'national' actor, as distinct from an 'international' one, obscures the vast differentiations relevant to any specific context. In Ethiopia, for example, highlander administrators and security personnel recruited and dispatched to oversee aid projects or other governance activities are perceived quite differently by lowlanders in SRS than those recruited from SRS itself; the former would certainly not be seen as 'local' by communities living in SRS, even if they are perceived as such by international organizations. The question of shared or hybridized worldviews and epistemologies (Mac Ginty 2011) is also poorly captured in international/national/local distinctions, particularly given the fact, as Autesserre (2014) and Heathershaw (2016) argue, that the frames of reference and habitus for those working in intervention sites can be more meaningful to them than national or organizational origins. 'Local' terminologies have also been critiqued for their romanticization of non-Western epistemologies and processes and for their normative, as opposed to conceptual, dimensions (Randazzo 2016).
Scholars have attempted to resolve some of these ambiguities through qualifying some of these terms (e.g. Oliver Richmond's local-local and local-international) or through suggesting alternative categories of actor altogether (Heathersaw 2016;Richmond 2011). It remains the case, however, in this author's view, that a language that can recognize some of the important distinctions mentioned above at the level of abstraction required to render broader conclusions on the relationships between intervention and power has yet to be developed-and is perhaps impossible to develop satisfactorily. This article will therefore use specific terminology where possible but will use the term 'internationals' to refer to individuals employed by a multilateral or bilateral donor agency without consideration of their nationality as well as those Western employees of INGOs and consultancy firms working on and in SRS. Civilians and communities in intervention contexts will be referred to as 'local' while state officials will be referred to as that, with their status as regional, national or federal employees specified where possible.
'Defensive living' between two worlds: bunkerization and intervention The recognition and conceptualization of the defensive and militarized world of contemporary international intervention emerged from two key areas of enquiry. The first originated in debates in the 1990s on 'human security' and the relationship between 'development' and 'security'-both in terms of their symbiotic conceptual link but also the implications of this for developmental interventions (Beall, Goodfellow, and Putzel 2006;Collier and Hoeffler 1999;Sen 1999;Stewart 2004). 9/11 and a range of military interventions by Western states-particularly in Afghanistan and Iraq but also Sierra Leonenevertheless re-shaped this debate around the question of how far developmental resources were being 'hijacked' for military and defence purposes by Western donors, INGOs and others (Albrecht and Jackson 2015;Woods 2005).
The ever more blurred line between development project and security enterprise in the post-9/11 context has been the subject of heated debate among scholars and practitioners -not least because of a growing focus by the UN and Western governments on 're-constructing' or, at least, 'stabilizing' 'fragile states' (Mac Ginty 2012). For scholars such as Duffield, this reflects not just a cynical commandeering of Northern policy agendas by security cliques but forms part of a much wider and insidious attempt by neo-liberal Northern elites to regulate and govern the everyday lives of those in the South (Duffield 2001a(Duffield , 2001b(Duffield , 2007. This critical commentary on the 'securitization of development'-which continues to rage-has increasingly come to focus on the spatial dimensions of intervention and aid delivery. The provision of aid supplies by Western security personnel in Afghanistan, Kenya and elsewhere, for example, have been viewed as a conspicuous indication of aid securitization and militarization (Bradbury and Kleinman 2009;Fishstein andWilder 2011 andSaferworld 2011). Scholars such as Lisa Smirl, however, have sought to deepen and develop this critique from a more sociological perspective, examining the militarized and separated architectures which house contemporary aid workers and help facilitate their interactions with 'locals' (Bliesemann de Guevara 2016; Smirl 2008Smirl , 2015Smirl , 2016. The sociology of international aid workers, or peacebuilders/peacekeepers, has also been the focus of a second body of work where the concern has been to analyse how international interveners view and approach their work and how they experience interaction with proximate Southern communities. For many, entry into this area of enquiry has come through critique of international actors who are unable to shift from a Western-centred, 'blueprint' worldview to one which takes account of local perspectives and priorities (Autesserre 2010;Lemay-Hébert 2009;Mac Ginty 2008). For others, the deleterious impact of 'everyday' practices of interveners-culminating in the creation of 'peacekeeping economies'-have been of central importance (Aning and Edu-Afful 2013;Edu-Afful and Aning 2015;Higate and Henry 2004;Jennings 2014Jennings , 2015Jennings and Bøås 2015).
Critical to this article, though, is the way in which this latter literature in particular has introduced a coherent concept of the international interventionary community. The delineation of a separate 'peacekeeping culture' has not only highlighted the architectural and spatial barriers and divisions erected by internationals, but the whole 'lifeworld' of this community-including how they construct, often erroneous, understandings of societies they engage with (Autesserre 2010(Autesserre , 2014Rossi 2006). Analysts depict a group whose everyday living is calibrated to limit contact with the 'local' for fear of attack, disease or other risk, while mind-sets engendered and preserved are those of separateness, privilege and difference (Autesserre 2014;Cain, Postlewait, and Thomson 2006). Duffield's analysis of the fortified aid compound brings these strands of literature togetherwith the compound standing not only for bunkerized space but bunkerized lifestyles.
It is important, however, to clarify the two dimensions of bunkerization that Duffield and others have developed: militarization and remoteness. The compound constitutes part of the former, which includes reinforced vehicles and security protocols. It also incorporates, however, the use of security and military forces to distribute or implement development or humanitarian goods-or, at least, to oversee these processes and 'protect' their providers. Thus one can see bunkerization in contexts where militarized compounds are less present, or wholly absent-such as eastern Chad, eastern Ethiopia or northern Kenya.
The other side of bunkerization, however, is 'remoteness'-the physical withdrawal of international personnel from key sites. On the one hand, this involves internationals' retreat to capital cities (sometimes in other countries) and engagement in periodic, brief 'fly-in' visits to areas in question. Throughout the 1990s, for example, many internationals dealing with war-torn northern Uganda based themselves in Kampala, while Nairobi became the home for many of their counterparts engaged 'in' Somalia-as Dubai and Jordan have become for internationals focused upon Iraq and Syria. This clearly has implications for the types of knowledge produced in Western diplomatic and developmental circles on regions beyond the metropolis (Fisher 2014;Roberts 2013).
On the other hand, remoteness entails the growing transfer of responsibility for undertaking, managing and evaluating internationally funded interventions by national staff and NGOs that are rarely afforded the same security resources as their Northern counterparts. This perhaps affords these actors a greater degree of agency in terms of implementation-and, indeed, shaping knowledge production processes-than might be possible under a more directly managed arrangement (Lewis and Mosse 2006). It also, however, transfers security risks identified by internationals to the local level (Shaw 2005) and raises difficult questions about whether international interventions are being implemented as intended or even if they are informed by any real sense of community concerns on the ground (Sandstrom 2014).
The inclusion of remoteness within our understanding of bunkerization is important for addressing one of the more central criticisms of the concept: its UN-centricity. Duffield, Autesserre, Henry and others focus heavily on UN peacekeepers and peacebuilders in their analyses. Their analyses therefore potentially aptly characterize phenomena in peacekeeping intervention sites such as Afghanistan, Somalia and South Sudan where the UN is the dominant international actor, but not in regions where the UN role is less central, such as parts of Ethiopia, Chad and Uganda. 2 Incorporating remoteness into bunkerization, however, allows us to apply the concept to many other interventionary contexts and organizations-particularly those where UN norms and protocols are less stringently observed and where multiple, mid-sized interventionary enterprises are the norm.
Where bunkerization's analytical purchase remains somewhat weak, however, is in its explanation of the drivers of the phenomenon. For most commentators, it is the initiative and agency of international actors and organizations that underlies the practice and mentality. While contestation of the material change in risk encountered by internationals in recent times forms a prominent part of this literature, broad agreement exists on the notion that internationals have deliberately driven the practice in reaction to perceived risk. This also appears to be the view of many practitioners themselves; Felix da Costa cites one INGO official in South Sudan's observation that 'bunkerization is essentially led by donors ' (2012, 7).
There is good reason, however, to query the decisiveness of this characterization. For example, a range of literature has recently explored the extent to which access and security is 'negotiated' between internationals and governments in many parts of the South, with security escorts and risk management plans imposed by the latter on internationals in an effort-in some cases-to strategically manage how Western knowledge on those states is produced (Del Valle and Healy 2013;Fisher 2014Fisher , 2015Harvey 2013;Magone, Neuman, and Weissman 2012;Pottier 2002).
More broadly, many scholars have highlighted the agency of African states in the international system in a number of contexts, underlining the extent to which apparent relationships of dependency and domination are in fact imbued with opportunities for securing agency on both sides (Beswick 2010;Brown and Harman 2011;Fisher 2012;Whitfield 2009). Mac Ginty, Richmond and other proponents of the 'local turn' in peace studies have also convincingly critiqued the dominance of Northern epistemologies and failed 'liberal' frameworks in the prosecution of peacebuilding initiatives by internationals. Instead, they have argued for a critical unpacking of local or indigenous agency in the emergence of sustainable peace practices in divided societies-from both a normative and a pragmatic perspective (Mac Ginty 2008Mac Ginty and Richmond 2013). These findings alone merit a more open approach to be taken to the question of what promotes and sustains bunkerization practices in many parts of the developing world.
Moreover, much of what is highlighted on the architectural, operational and practical side of bunkerization resembles not only contemporary practices of governance and state consolidation in parts of the South but also much more historic ones-in some cases practices which have taken place since the pre-colonial era. The remainder of this article will therefore seek to re-frame the discussion on bunkerization by 'bringing the state in'. This will be undertaken firstly through the drawing of parallels between historic statebuilding practices and contemporary bunkerization, followed by an historically informed analysis of the drivers of bunkerization in Ethiopia's SRS.

Genealogies of bunkerization in central-eastern Africa
In recent decades, many of the most prominent humanitarian crises-and focuses of international intervention in various forms-have been located at the peripheries of central/ eastern African states. War in northern Uganda, genocide in Darfur, refugee crises in eastern Chad, conflict in the Ogaden, civil strife in southern Sudan-borderlands have been a focal point for suffering, atrocities, insurgency and counter-insurgency for much of the recent past and thus have become the key sites for intervention, directly or otherwise.
A key point to make in this regard is that these regions' entry into the modern global system of states-and their subsequent governance-strongly reflect phenomena discussed and critiqued in the bunkerization literature. Though polities and systems of governance existed in all the examples cited above prior to the nineteenth century, the set of processes which led to their incorporation into the modern states of Uganda, Sudan 3 , Chad and Ethiopia and the forms of engagement they have continued to have with governments in Kampala, Khartoum, N'Djamena and Addis Ababa to this day closely resemble both militarization and remoteness, as discussed above. That is to say, statemaking and statebuilding in the peripheries of these four states has long been associated with the practices conceptualized as novel and Western-led in much contemporary intervention literature.
The incorporation of parts of Darfur, western Ethiopia and southern Sudan into what eventually became (largely) Sudan by Turco-Egyptian forces (with British support) in the nineteenth century, for example, was undertaken violently and maintained through crude and exploitative military rule (Leonardi 2013;Gonzalez-Ruibal 2011). Power was centred around fortified garrisons or zaribas occupied by Turco-Egyptian commanders and their retainers. Leonardi (2013) has vividly reconstructed the environments of these 'clearly demarcated and exclusionary' structures and the local economies which grew up around them-structures and processes which resemble, from this perspective, Duffield's aid compounds and the development of 'peacekeeping economies' (Higate and Henry 2014;Jennings and Bøås 2015).
Importantly, post-colonial Sudanese governments maintained a similar approach to ruling the south and west-northern commanders being the main representatives of the state-housed in military bases to govern and, where necessary, subdue the population (Rolandsen and Leonardi 2014;Deng Kuol 2014). Even members of the rebel Southern People's Liberation Movement (SPLM)-now the ruling party of South Sudan -have reflected on their militarized approach to engaging with and governing civilians in occupied territory they held between the 1980s and 2000s. 4 Similar observations can be made on the incorporation of modern SRS into Ethiopia. The region-populated primarily by pastoralist communities-was violently subdued by the imperial Ethiopian army in the later nineteenth century and left largely 'ungoverned' by the centre save for periodic campaigns led by distant military officers stationed in urban posts and fortified military structures (Hagmann 2014, 14;Markakis 2011, 134-148). The presence of the state in the region today does not look dissimilar.
Faced with a separatist insurgency since the mid-1990s, the Ethiopian government has, in SRS expert Tobias Hagmann's words, 'returned to [the] garrison rule' which characterized its presence in SRS for much of the twentieth century (Hagmann 2013). Indeed, Hagmann and Korf characterize successive Ethiopian regimes' approaches to SRS in terms of Carl Schmitt's and Giorgio Agamben's 'state of exception', noting that 'physical violence meted out by subsequent Ethiopian governments has been a constant in the past 120 years' (Hagmann and Korf 2012, 209). In illustrating this point, Hagmann refers to topographic maps of the Somali region produced by Ethiopia's official Mapping Agency during the 1970s 'which depict major towns and settlements within the Ogaden as "military camps" with no reference made to the existing (Somali) names of these locations' (Hagmann 2005a, 532, note 6).
North-western Kenya, northern and north-eastern Uganda and eastern Chad were also violently incorporated into their modern states by a combination of European and Turco-Egyptian colonial projects and have continued to be governed as enemy territory, to some degree, by state officials (Knighton 2003). David Anderson, for example, traces the continuities between the militarized character of British colonial rule in northern Kenya -'always treated as a "special district" under a system of military administration … garrison government'-and the post-independence regimes of Jomo Kenyatta (1963-78) and Daniel arap Moi (1978 (Anderson 2014, 660).
Perhaps most startling in this regard, though, was the militarization of much of northern Uganda during the 1990s and the forced displacement and encampment of millions of civilians by the Ugandan army in its war against the Lord's Resistance Army (Dolan 2009). During this period, the state and the military were once again synonymous for many civilians; the president, Yoweri Museveni, temporarily moved the government to the region during the mid-1990s, not to oversee the humanitarian operation but 'to command the army there himself'. 5 Remoteness has also been a central characteristic of state 'presence' in these regions since the nineteenth century. Thus in nineteenth/early twentieth century eastern Ethiopia and Darfur, local militias and auxiliaries were often employed as proxies to maintain state control by distant commanders-not unlike Khartoum's approach to janjawiid militias in Darfur during the 2000s and Addis Ababa's use of the liyu militia in SRS. Indeed, as Øystein Rolandsen notes, 'throughout the history of Sudan as a state, autonomous armed groups have been employed as slave raiders, auxiliary troops, border police and … tools in government counterinsurgency campaigns … employing militias has become a standard procedure for successive regimes in Khartoum' (2007,165).
More often, though, the state came to maintain its rule through relationships of (unequal) mutual dependency with local leaders (actual or appointed by state officials), particularly chiefs in southern Sudan/northern Uganda and clan elders in Somali-inhabited regions (Amone and Muura 2014). These selectively chosen actors became the voice and arm of the state in many of these peripheral regions owing to their perceived ability to mobilize support and to 'translate' the demands of government to distant communities which administrators were unwilling to reach out to themselves (Markakis 2011).
Significantly, though, these leaders were valuable because they could speak the language of their colonial or proto-colonial overlords (both literally and metaphorically)-something which was not a coincidence. Leonardi (2013) has shown, for example, how southern Sudan's pre-colonial community leaders transformed into chiefs during their growing encounters with various manifestations of the state (hakuma) during the nineteenth and twentieth centuries. Over time, these leaders came to understand and appreciate how to most fruitfully position themselves vis-à-vis state representatives to maximize their own agency as de facto rulers at the local level. In so doing, they played an important and strategic role in managing the production of knowledge by both the hakuma and its citizens.
This emerging class of brokers between 'life worlds' has not simply been a prominent feature of twentieth century governance in parts of southern Sudan, Uganda and Somalia but of the modern development enterprise (Mosse and Lewis 2006;Rossi 2006). A key feature of bunkerization has been the transfer of responsibility for administration and operations to national staff and organizations. The selection of such 'local' brokers by international donors and INGOs, though, has been premised on a similar rationale to those of colonial and proto-colonial administrators: perceived ability to mobilize and represent communities, but familiarity with the world of the international. 6 Moreover, as Clare Paine has demonstrated in the case of northern Uganda, for example, for some modern brokers an attempt to fit the profile of those of old ones-in this case the 'paramount chief'-has sometimes been a key strategy for building credibility with, and securing resources from, international interveners (Paine 2015).
The intention of this overview is not to suggest that contemporary bunkerization practices are simply an unconscious extension of historic statebuilding enterprises. Nor is it to imply that international intervention is heavily informed by the same militaristic, imperialist logics as the European, Turco-Egyptian and Ethiopian colonizers and expansionists highlighted above. The drawing of these parallels nevertheless underscores the enduring character of the state in many central/eastern African peripheries. That is to say, forms of engagement and interaction between state authority and local communities in many of these regions today continues to reflect the same logics of militarization and remoteness that informed initial encounters in the nineteenth century.
This is a crucial lens through which to view the practices of modern international interveners, since development interventions do not occur unilaterally-they must, at the very least, come to some form of accommodation with the formal state authorities. In some cases this is a question of registration or permission, while in others a more regulated and continued negotiation of access is required. This applies particularly to UN and international donor agencies but also to INGOs, whose de-registration and expulsion by state authorities has occurred, and been observed, in Darfur, SRS and elsewhere in the last decade.
The point, then, is that intervention does not occur in a vacuum-the state may be largely absent in the borderland but is still recognized, and usually referred to, as the presiding authority and determiner of movement and operation by internationals. The extent to which states in eastern/central Africa have shaped and managed the character of international interventionary presence in line with a persistent historical modus operandi of militarized encounters and remote governance is therefore a central, and largely unexplored, question. The degree to which interventions undergird these deleterious forms of rule in the African periphery-albeit unconsciously-is also a crucial normative concern that arises herein. The remainder of this paper will explore this in the context of Ethiopia's SRS region.

Internationals in SRS: a brief overview
Contemporary Ethiopia is administratively a federal state divided, since 1991, into nine regional states. Under the current Ethiopian People's Revolutionary Democratic Front (EPRDF) ruling coalition, regional states have been created along ethnic lines as a means to break links with previous regimes perceived to have been highly centralized, dominated by an Amhara elite and unrepresentative of ethnic groups outside the highland 'core' of the country (Abbink 2011).
One such state is SRS in the eastern part of Ethiopia, one of the largest in the country in terms of both area and population. Annexed to Ethiopia by the former Abyssinian empire in the late nineteenth century, the region once known as the 'Ogaden' has remained on the margins of the Ethiopian polity both physically and politically ever since. Though governed by a native Somali administration in the town of Jijiga theoretically elected by Somali Ethiopians, the region's rulers have in fact largely been the de facto appointees of the increasingly authoritarian federal government in Addis Ababa via its Ministry of Federal Affairs (MoFedA) since the 1990s and, indeed, (largely northern Tigrayan) Ethiopian military commanders (Hagmann 2005a(Hagmann , 516-517, 522, 2005b(Hagmann , 2014Khalif and Doornbos 2002). This is not to say that regional officials have not been able to carve out agency in their relations with their federal counterparts, though it is clear that their authority derives from above rather than from below. SRS remains one of the poorest parts of Ethiopia and has received minimal investment from the federal government in terms of infrastructure and other forms of support. Economically, the region is largely dominated by pastoralists-a way of life federal officials, like their predecessors under the socialist Derg (1974-91) and imperial regimes (overthrown in 1974), find difficult to understand, leading to the latter's support for a range of clumsy and deleterious 'villagization' and sedentarization programmes in the area (Hagmann and Mulugeta 2008). This perspective has fed into a broader view among Ethiopia's highland rulers of SRS as a terra nullis requiring subjugation and civilizing (Clapham 2002).
The marginalization and under-development of the region has also been exacerbated by conflict in recent decades, with the secessionist Ogaden National Liberation Front (ONLF)-ruling party of SRS between 1992 and 1994-having waged a violent separatist campaign since leaving the EPRDF coalition in the early 1990s. The ONLF insurgency has been met with equal brutality by the Ethiopian military and the regional government'sinitially informalized-special forces, the liyu police.
SRS has also largely been among Ethiopia's more neglected regions for international donors. This lack of interest has nevertheless begun to change since a major drought and famine struck the wider region between 2011 and 2012. Recent years have seen a number of states establish-or explore the possibility of establishing-longer-term development interventions in this part of Ethiopia. Most notably, the UK's Department for International Development (DFID) launched a multi-million pound 'Peace and Development Programme' (PDP) in the early 2010s (initially as an inception mission, though operational in part since 2013) focused on four components-service delivery, fostering of greater professionalism and accountability in the security and justice sectors, assisting in the creation of livelihood opportunities for pastoralist communities, and improving governance. DFID operations are managed by a consortium of INGOs and NGOs under Save the Children, including the International Rescue Committee, Islamic Relief, Ogaden Welfare and Development Association, Mercy Corps, Oxfam and ZOA (DFID 2013). Since 2013, the German and Swiss development agencies (Deutsche Gesellschaft fur Internationale Zusammenarbeit (GIZ) and Swiss Agency for Development and Cooperation (SDC)) have begun to plan for, and trial, interventions focused around capacity-building for drought resilience among Somali-Ethiopian communities, while the US-Ethiopia's leading bilateral donor, but a relatively minor player in SRS-has primarily focused on food and water security programmes. 7 A number of UN relief agencies operate in the region, particularly focusing around refugee camps, notably the World Food Programme (WFP) which is housed in defensive compounds in Gode, Dollo Ado and Degehabur. United Nations High Commission for Refugees (UNHCR) also has an office in Jijiga-the regional capital-and the UN runs a small 'guest house' there for visiting international staff (UNDSS Ethiopia 2014). Though UN security advisers and liaisons are stationed in a number of these locations, they are invariably 'national' rather than international in origin. Finally, the World Bank has run a community-based development programme for the region's pastoralists (the Pastoral Community Development Project, PCDP) since 2003. Most of these programmes are implemented by the federal or regional government or by the many humanitarian INGOs and NGOs in the region, most notably Mercy Corps, Oxfam, Save the Children and ZOA together with a range of consultancy groups, implementing agencies and security outfits. 8

Drivers of bunkerization: the role of the Ethiopian state
The difficulties of accessing many parts of SRS together with the perceived high level of insecurity there arguably render it an ideal candidate for bunkerization behaviour and mentalities among international actors. Indeed, at one level there appears to be a high level of militarization and remoteness in the character of international interactions with civilians, communities and organizations in the region.
Few internationals are permanently based in SRS, and visits by international personnel are often heavily restricted by a donor/agency headquarters for both security and practical reasons. Indeed, according to interviewees, brief trips beyond Jijiga are often made only with Ethiopian military escorts and equipment, or with UN logistical support-often in the form of reinforced vehicles or helicopters. 9 Several donors, notably the World Bank, rely heavily on UN security recommendations and advice with regard to sending internationals into the area even for short trips; 10 the UN Department of Safety and Security (UNDSS) provides a weekly précis of security 'incidents' across the country for interested actors and advises internationals, and a number of other communities, on the perceived relative risks of travelling to particular areas. The proposed German-Swiss intervention noted above was to proceed only after the completion of a risk assessment exercise undertaken by an external consultant-whose report, which was in the process of being written at the time of the fieldwork, sets the parameters for this initiative-and, indeed, was to define GIZ and SDC's understanding of the SRS context. 11 Moreover, the UN presence in the region takes its most prominent form in militarized compounds; bunkered structures surrounded by barbed wire separating local and international space. For international consultants and NGO personnel working in the region (usually under the commission of the UN, an international donor or an INGO), UNDSS or other institutional security advice has often compelled these actors to remain behind compound walls-particularly in Gode-for much (if not all, as one consultant interviewed noted) of a trip to the region. 12 In terms of operations, international actors rely heavily upon their Ethiopian government counterparts, both civilian and military, in some contexts to deliver aspects of their programmes in the region. A 2010 review of development interventions in the region, for example, noted that WFP 'hands over all relief food to the Government of Ethiopia at the time of its arrival … in the country [for distribution]' (DAG Ethiopia 2010, 22).
International encounters are also characterized by a strong degree of remoteness. Visits by internationals to the region are usually brief and heavily circumscribed. Current and former diplomats, consultants and humanitarian personnel have described how their trips have largely focused around meetings with regional state political and judicial officials (often in Jijiga) as well as with diaspora returnees from Europe and representatives of INGOs and implementing agencies tasked with delivering donor programmes. Encounters with 'locals' who are not part of the state elite (federal or regional) or the development community are therefore extremely rare.
International donor-funded interventions are also largely implemented, and often evaluated, by increasingly distant chains of consortia and implementing agencies, often run and managed by local NGOs. On the one hand, this increases room for local agency in the management of these programmes. On the other hand, however, it expands the physical, emotional and epistemological gap between the life worlds of internationals and locals. This leads to an arguably circular and restricted production of knowledge. In a September 2013 review of the nascent PDP, for example, DFID officials did not incorporate a field visit in their evaluation, but instead held meetings with implementing partners and representatives of the Ethiopian state to inform their analysis (DFID Ethiopia 2013, 8).
This was considered justifiable given the very early stage of the PDP's development and operationalization at that time. The episode nonetheless reflects the more general character of international involvement in the region, where chains of organizations not only help design and implement interventions but also represent the primary (sometimes sole) sources of information on the region and its peoples for their international interlocutors. This has both epistemological and practical consequences, as Sandstrom (2014) has similarly noted in relation to internationals' 'remoteness' in Afghanistan.
One should, of course, be careful about essentializing international engagement in SRS. Security and visit protocols, risk assessment guidelines and training requirements vary considerably across headquarters and are rarely observed to the letter by international staff who travel to the region in person. Likewise, it was clear from interviews that many of the latter have sought to resist or work around such HQ-imposed restrictions and that a number relied more on their own extensive experience of working in SRSand networks developed therein-to ensure their own safety. 13 As Felix da Costa and  have noted in relation to South Sudan, there is also some evidence of internationals seeking to establish 'ways to stay' in the region when this is against the broader security protocols.
The provenance of this general bunkerization behaviour, however, is open to question. For while HQ diktats and a generalized culture of risk clearly shape the form of international intervention in SRS for many interviewees, the role of the Ethiopian state was more often highlighted. Attempts to establish a permanent, non-militarized footing in the region by one major donor in recent years, for example, were rebuffed by Addis Ababa. 14 A range of personnel from a variety of organizations also noted that Ethiopian military escorts were in fact imposed sine qua non for many internationals seeking to travel beyond Jijiga-and that permission to do so would be denied or vehicles turned around at roadblocks otherwise. 15 Moreover, many interviewees claimed that building the Ethiopian state (sometimes including, usually indirectly, the military) into the implementation of interventions was, in fact, often a prerequisite for the state's granting of approval for the programme to take place. Internationals universally noted that their interventions had to be fully 'aligned' with the Government of Ethiopia's (GoE) National Development Plan in order to be approved and permitted to operate. 16 Most interviewees could also recall at least one occasion during their tenure when the GoE had 'refused' project or programme aid or expelled a colleague when the interventions proposed or part-implemented had not fitted into the Plan. 17 Many stressed that their interventions were allowed to take place only through the existing state apparatus, with a range of state coordination bodies, notably MoFedA and the Bureau of Pastoral Affairs. 18 Moreover, one donor official noted that his programme had needed to agree with the GoE beforehand which woredas (the smallest unit of local government in Ethiopia) the project would operate in with a range being 'jointly agreed' as too dangerous to enter. 19 The Ethiopian state's management of space which internationals can access in SRS has been a central and longstanding characteristic of its engagement with this community and represents, in part, a strategic attempt to prevent the production of a non-EPRDF-sanctioned knowledge on the insurgency and counter-insurgency. 20 State officials heavily restricted humanitarian access to parts of SRS at the height of the ONLF insurgency between 2007 and 2008 and expelled a number of organizations viewed as assisting the group (including Medecins sans Frontiers, MSF) (Binet et al. 2011). Many have only recently returned at the time of writing. In the name of balance and providing a broader context, it should be noted that UN and other internationals also became the targets of increased numbers of kidnappings, killings and other violence by the ONLF during this period (Powell 2007; Voice of America 2011).
At the time of fieldwork fewer restrictions existed, although respondents emphasized the non-negotiability of international access to parts of SRS cited as international no-go areas by officials in Jijiga and Addis Ababa. Both the PDP and the World Bank's PDCP, for example, were allowed to proceed only in areas specifically agreed with state officials with certain proposed locations 'bluntly' refused by Ethiopian state interlocutors. Addis Ababa has also imposed strict thematic limits on donor involvement-with the Safety and Justice element of DFID's PDP being suspended a fortnight into its design phase owing to the perceived sensitivity of the topic. 21 One senior official opined that 'the security services saying no' represented the biggest impediment to access in most of the region while several consultants and donor officials argued that they had been able to move more freely in Afghanistan and Somalia than in much of SRS for similar reasons. 22 Importantly, the GoE's management of international space has also been complemented by an attempt to manage network building by internationals-and thus 'sources' of knowledge. Virtually all respondents described the complexities of building links in the region beyond groups and actors aligned with the military or regional/ federal government. A particular concern in this regard was the military and political establishment's frequent flagging of particular communities as being linked to, or sympathizers of, the ONLF and thus 'out of bounds' to internationals. Many respondents noted that the vagaries of this messaging discouraged internationals from seeking to establish unmediated relationships with those in SRS. 23 Several international officials highlighted the frequent insistence of the regional government that its advisers accompany them on any trips in the State and that only its approved staff, including interpreters, be used. Indeed, one official noted that an attempt to add an additional interpreter to the security escort selected by regional officials on one such trip led to a heated and sustained disagreement. 24 To some extent, this promotion of more militarized and remote encounters between internationals and locals by the Ethiopian authorities is a consequence of several aspects of Ethiopian political culture. Firstly, successive Ethiopian regimes have operated within a context of institutionalized secrecy, suspicion of external actors and reluctance to share information unnecessarily (Hansen 2006;Vaughan and Tronvoll 2003). This has been augmented under the current EPRDF regime, which came to power largely without international support and emerged from a disciplined and necessarily secretive guerrilla movement.
Moreover, the EPRDF has relied heavily upon international donor support since early in its tenure and has worked hard to limit international exposure of issues, such as counter-insurgency tactics in SRS, which might imperil this assistance (Fisher 2013). Finally, the EPRDF has adopted a resolutely independent approach to domestic policy -promoting a coherent and comprehensive vision for a top-down 'developmental state'. 25 Within this dispensation, development policy has been viewed as something which should be largely state-owned and which international actors should either align with or take their funding elsewhere (Dereje 2011;Furtado and Smith 2009). This focus on ownership and independence-dubbed 'Ethiopian exceptionalism' by one interviewee 26 can partly be explained by the ideological robustness of the EPRDF regime and its leadership (at least until 2012) coupled with Ethiopia's history as one of Africa's only states not to have been colonized (the brief Italian occupation during the Second World War notwithstanding). Cultural suspicion of the intentions of foreign powers within the Ethiopian elite also contribute to this way of thinking of and engaging with the outside world; as one Ethiopian official argued, 'in Africa, the problem is that if you are weak you will be manipulated'. 27 One cannot, however, discount the influence of historical practices of statebuilding by Ethiopian administrations in the region now known as SRS on contemporary practice. For while the EPRDF's overall approach to statebuilding differs markedly from its Marxist and imperial predecessors, engagement in the Ogaden differs little (Hagmann and Korf 2012). Indeed, as one SRS state official suggested, 'the main difference for many in the region under EPRDF is that the soldiers are Tigrayan not Amhara'-a reference to the initial reconstitution of the Ethiopian military around the Tigrayan TPLF under the EPRDF, and the continued prominence of TPLF figures in the higher ranks of the military in particular. 28 As Hagmann notes, 'successive Ethiopian regimes perceived government action in the Somali Region primarily in military rather than political terms' (Hagmann 2005a, 512).
In this context, it is possible to link Ethiopian state approaches to governing SRS to bunkerization behaviour among international interveners. For Ethiopian officials, the modus operandi of engaging the region has historically focused around militarized and distant encounters and this has been the behaviour promoted by Addis Ababa and Jijiga among external actors. That is to say, the bunkerization of aid in this region has not simply been a donor-driven phenomenon. International bunkerization in SRS is instead most usefully seen as an historical form of engagement between state actors and local communities now reproduced among external interveners at the instigation of their Ethiopian state hosts.
Concluding thoughts: sovereignty, peripheries and statebuilding in the era of bunkerization When presenting this article's argument in a range of settings in Africa, Europe and North America during 2014-16, one qualification suggested by several thoughtful respondents focused around the strength or hardness of the state concerned. Thus, where state drivers of international bunkerization could be convincingly argued for in strong states -such as Ethiopia, Rwanda or Sri Lanka-the story was less persuasive in the case of weaker ones.
While accepting the different degrees of influence state actors have on the behaviour of internationals in their territories, this article has nonetheless sought to construct its argument less around the strength or weakness of a state and more around how peripheries and borderlands have been incorporated into these states. It has then explored how external relations with local communities during and since this period of incorporation have come to be reproduced by successive administrations and international interveners. This focus on borderlands allows for the incorporation of traditionally 'strong' states such as Ethiopia into the analysis alongside weaker ones such as Sudan and Chad. Moreover, this approach acknowledges the growing ambivalence among Africanist scholars on the degree of choice exercised by African states (including inter alia Uganda, DRC and South Sudan) in projecting strong or weak state presence in parts of their territories (Fisher 2014;Schomerus and de Vries 2014).
In highlighting the agency of African states in driving the bunkerization process the study does not, of course, deny the prominence and provenance of Western actors, mind-sets and norms-some of which can indeed be argued to flow from colonial mentalities and structurings of space, sovereignty and the 'other'. Nor does it seek to deny the complex interactions between historical patterns of behaviour and engagement and the conscious, intended actions of individual state actors and groups in the present. The aim of the article has been to highlight the state role in a process so far understood primarily in terms of novelty and Western agency rather than to resolve the structure-agency debate.
The article's findings nonetheless raise broader, more challenging questions for the international intervention community. The Ethiopian state has not been alone in promoting bunkerized behaviour-Karlsrud and Felix da Costa (2013), for example, highlight the role of the Chadian state in imposing military escorts on humanitarian workers in its eastern periphery. Goodhand (2010) emphasizes the Sri Lankan government's successful politicization and securitization of the presence of humanitarian organizations in that country during, and since, its civil war. Indeed, state restrictions on the movement and remits of internationals have increased considerably in recent decades and many states, including Sudan, Rwanda, Chad, Eritrea, Ethiopia and Zimbabwe have become steadily more confident in their use of sovereignty norms to expel or circumvent the activities of external actors on their soil (Fisher 2015;Magone, Neuman, and Weissman 2012;Tull 2010). This raises questions not only about the plight of increasing numbers of communities cut off from humanitarian assistance but about how those in the policy and humanitarian world can access 'ground truth', in Duffield's words, on these communities-and thus even identify a humanitarian crisis in the first instance.
In the case of SRS, international respondents often discussed their approach to engaging in the region in terms of doing it 'their [the GoE's] way' but pushing 'little by little' for greater room or undertaking 'what we can get away with'. However, the quandary is where internationals should draw the line in terms of allowing their incorporation into statebuilding practices in parts of the South as vehicles for possible future greater humanitarian or developmental autonomy. A number of respondents, for example, felt uncomfortable about the perceived link between international intervention and the GoE's sedentarization programmes, while others highlighted recent accusations of proposed UK funding for liyu paramilitaries in SRS (Quinn 2013; see above).
It is increasingly apparent that a range of polities now view the international system as something to be incorporated into a broader process of-often semi-authoritarian-statebuilding rather than a set of ad hoc actors to be negotiated with bilaterally and periodically (Fisher and Anderson 2015). Across eastern and central Africa particularly, many states no longer permit interveners the 'choice' of supporting them or engaging in other activities outside of their structures. This trajectory is not without its advantages of course-greater humanitarian and developmental coordination can be a positive thing even in semi-authoritarian states, depending on the commitment of these states' governments to delivering humanitarian and developmental goods and resources. Moreover, from a normative perspective, one could argue that unsupervised and untrammelled interventions by international actors is anachronistic, neo-colonial and should, indeed, be a thing of the past.
Ultimately, though, this article has not linked bunkerization and remoteness to the empowerment of communities and citizens in Africa but to longstanding processes of militarized governance and statebuilding by national elites and their colonial predecessors. Many of the contemporary governments being referred to in this regard are building their states around rent-seeking, particularism, exclusion and oppression, delivering macroeconomic growth, in many cases, but distributing the dividends quite unequally (Jones, Soares de Oliveira, and Verhoeven 2013). International actors working in these countries and regions must therefore increasingly face up to the fact that their interventions form part of a broader enterprise-the creation of illiberal and ultimately deeply unstable authoritarian states.

Notes
Notes on contributor Jonathan Fisher is a Senior Lecturer in the International Development Department, University of Birmingham. His work focuses on: (1) exploring the place of African states in the international system and the role and agency of these states in managing their relationships with, especially, Western donors and institutions; and (2) the production of knowledge on African states and societies, particularly in relation to conflict and security. He is particularly interested in Eastern Africa and is the PI on an ESRC Seminar Series on 'From Data to Knowledge: Understanding Peace and Conflict from Afar' (ES/N008367/1) and an AHRC project on 'Witchcraft and Conflict: Exploring Alternative Discourses of Insecurity' (j.fisher@bham.ac.uk).