Janice Gross Stein 



        The challenges faced by non-governmental organizations seeking to mitigate violence within the context of “complex humanitarian emergencies” create new dilemmas and require new strategies. These emergencies arise from violence inflicted by one group against another within the confines of a state, from the capture of state institutions by one group, or by the collapse of these institutions and the failure of governance. They develop within a context of disengagement by the major powers and the privatization of emergency assistance.

        I first analyze the dimensions of complex humanitarian emergencies, define the dilemmas humanitarian NGOs face and their implications for conflict resolution, and examine the changing international context to establish the scope of disengagement and privatization. I then assess the troubling evidence that humanitarian NGOs have contributed inadvertently to the escalation of violence rather than to conflict resolution. I explore three possible strategies, some of them counterintuitive, which could contribute to the mitigation of the violence and to conflict resolution.


The Growth of International Non-Government Organizations 

        In the examination of the prevention, management, and resolution of violent conflict, the role of international non-governmental organizations (NGOs) traditionally has received little more than a footnote. Analysis of the major powers, regional organizations, and the United Nations dominated the discussion. In the last decade, however, particularly since the end of the Cold War, non-governmental organizations have become more prominent -- and more controversial -- especially in the complex humanitarian emergencies that arise from local conflict. They are more important for two reasons: the number and importance of NGOs has multiplied exponentially and the spectrum of conflict, which is the focus of international attention, has broadened.

        At least three important changes have occurred in the profile of international non-governmental organizations. First, there has been a significant increase in the numbers of international NGOs working in the south, at the same time as governments in the north have privatized their assistance programs (Smith and Lipsky, 1993; Gordenker and Weiss,1996). The major powers have increased their funding to NGOs even as they reduced their spending on bilateral assistance programs. From 1980 to 1993, the number of NGOs in the north focused on development almost doubled. This growth is a direct outcome of the restructuring of the state and welfare systems by northern donors during the 1980s. In some countries, official development assistance has effectively been privatized.

        International institutions have also vastly increased the proportion of their funding that they channel through NGOs. The European Commission, for example, raised its funding for NGOs from zero to 40 percent, with a corresponding reduction in bilateral emergency aid from 95 to 6 percent between 1976 and 1990. Even in a short period of three years, between 1988/89 and 1991/92, the proportion of development assistance channeled through NGOs by the Department for International Development (DFID) in Britain increased by 28 per cent. The growing importance of NGOs as international actors is a function both of the privatization of assistance and the withdrawal of states and international organizations from the field. Increasingly, it is NGO personnel who are providing relief and assistance to the victims of conflict in the space vacated by states and international institutions. This assistance has become embedded, however, in the larger context of violent conflict, at times, with unanticipated consequences.

        Not only have the numbers of NGOs increased, but also new kinds of NGOs have developed. A decade ago, it was largely non-governmental organizations with religious affiliations that focused on mediation and conflict resolution. Now, secular NGOs, specializing in conflict prevention and resolution, and operating independently of states and the United Nations, are active in the field. International Alert and the International Crisis Group, two of the best known among these new non-governmental organizations, have played an especially important role. Although the resources of these NGOs who specialize in conflict resolution are minuscule compared to the NGOs who provide humanitarian relief, their political impact is often out of proportion to their size. Engaged in such activities as negotiating hostage release, supporting local NGOs committed to peace building and conflict resolution, advising parties to the conflict, and helping to facilitate political negotiations, the conflict resolution NGOs are an important part of the international political landscape. At times they complement and at times they compete with the traditional diplomatic efforts of the United Nations, regional organizations, or individual states.

        It is not only the newer non-governmental organizations that are committed to conflict prevention and resolution. Increasingly, the large, traditional development and relief NGOs have adopted components of the conflict resolution agenda in their emergency programming. Action Aid, for example, has explicitly designed programs for internally displaced persons around principles of reconciliation. This represents a significant departure for most of the large NGOs, and one which is likely to represent a growing trend in their activity, as political backing and funding for this conflict resolution activities increases. Conflict prevention and resolution are now squarely on the NGO agenda.

        The focus of this paper is not on the new NGOs who specialize in conflict prevention and  resolution. As important and innovative as they are, it is too early to assess their impact systematically (Voutira and Brown, 1995). Rather, I focus on the role of the large NGOs, committed to humanitarian assistance and relief, in the context of a “complex humanitarian emergency” that grows out of violent conflict.

        An examination of the role of the largest relief organizations in complex humanitarian emergencies illuminates some of the central dilemmas of conflict resolution. These NGOs, with long-standing commitment to a humanitarian ethic, now find themselves in the eye of the storm. They are the target of sustained criticism that their relief exacerbates and fuels conflict. Examination of this debate goes far beyond the role of humanitarian NGOs. It illuminates attributes of violent conflict in the post-Cold War system, the complexity of contemporary humanitarian emergencies that grow out of violent conflict, and, most importantly, the security vacuum which is creating these acute dilemmas for NGOs and impeding effective conflict resolution.

        I define a complex humanitarian emergency as a multi-dimensional humanitarian crisis that is created by interlinked political, military, and social factors, most often arising from violent internal wars that in turn frequently are the result of state failures. It almost always involves some combination of mass population movements, severe food insecurity, macro-economic collapse, and acute human rights violations up to and including genocidal projects.

        State failure can refer to a lack of capacity on the part of state institutions to secure territory, enforce authority, or maintain a monopoly on coercive violence. The state cannot secure the basic rights of citizens, fails to provide fundamental protection, and becomes unable to fulfill essential international legal responsibilities. As the authority and capacity of the state weakens, it may invite attack from disaffected segments of the population who can mobilize the resources. In response, a weakening state may attack its own population in an effort to reassert authority, or the state may collapse or implode. Collapse is a severe reduction in capacity, authority, security, identity, institutions, and, at times, territory, so that institutions effectively cease to function. It can be understood as the most severe form of state failure. The Somali bombing of sections of northern Somalia is an example of the former, while the flight of Siyaad Barre from Somalia is an example of failure through collapse.

        Alternatively, segments of the population can capture even a relatively strong state for parochial purposes and use theinstruments of the state to attack segments of the population. The militant Hutu militias, motivated by their strong opposition to a negotiated power-sharing agreement, itself the result of a major international effort at conflict resolution, captured the state in Rwanda in April 1994 and launched a genocidal massacre of Tutsis and moderate Hutus.

        Before examining the theoretical and policy controversies that are swirling around the role of NGOs, I briefly describe the cases and the evidence that is used to evaluate the competing claims current in the literature. 

Evidence and Case Selection 

        The study draws on three principal case studies as well as from ongoing tracking of other complex humanitarian operations in Africa. Somalia, Rwanda, and Sierra Leone are three of the best known cases where political violence led to a large-scale humanitarian disaster that required a multi-dimensional response. They are the principal case studies (Jones and Stein, 1999; Jones, Stein, and Bryans, 1999). Liberia and Burundi share some of these characteristics, and have been the location of important humanitarian programs; Eastern Zaire was the site of multi-faceted response to a complex emergency, and the focus of some of the most vociferous debates about policy responses. These three cases have been tracked, as important checks on evidence drawn from the principal cases.

        The cases were chosen at different points along the crisis time line: Sierra Leone, at the time a case of incipient state failure; Rwanda/Zaire, an on-going crisis; and Somalia, a post-emergency, in the aftermath of large-scale intervention. This variation in time line permits some consideration of competing theoretical propositions against different bodies of evidence. Restriction of the cases to Africa was deliberate. Once the Cold War ended, the attention Africa received from the major powers dropped precipitously.  As the major powers withdrew, and economic failure and violence increased, and, in some cases, states collapsed, development and humanitarian non-governmental organizations significantly increased their presence.   

The Dilemma in Context 

        In the last several years, humanitarian NGOs have increasingly found themselves trapped in an acute dilemma. This dilemma is best exemplified in the work that was done with Rwandan refugees in eastern Zaire, in the aftermath of the genocide and the victory of the Rwanda Patriotic Front in 1994. Agencies charged with running refugee camps, using the most tested and progressive methods of camp management, nevertheless found themselves by the autumn employing mass murders and war criminals as local staff. The perpetrators of the genocide had re-imposed authority over hundreds of thousands of refugees under the supervision of the United Nations and humanitarian NGOs, and were organizing to use the camps as a springboard to attack the government of Rwanda. Humanitarian assets were being used to fuel rather than resolve conflict. A more perverse outcome from the perspective of humanitarian NGOs is difficult to imagine.

        The perversion cannot be explained exclusively or even largely by flawed NGO practices. Certainly, practices were flawed at times and could be improved, but, in this case, better practices would not have prevented the militias from organizing the camps. The roots of the unanticipated and negative consequences of assistance are found, paradoxically, first in the humanitarian ethic, which informs the work of many of the large NGOs, and then in the attributes of contemporary civil war, the global security vacuum, and the changing political economy of international assistance.

        The humanitarian ethic. Humanitarian work is rooted in a charitable ethic, the imperative to come to the assistance of those in dire need. The essence of humanitarianism is its neutrality and its universality, its refusal to choose one distress over another (Kouchner, 1993; Hendrikson, 1998; Delmas, 1997:201.). Not only those NGOs who deliver relief assistance, but those working to facilitate development and conflict resolution seek to promote human welfare among distressed populations. Humanitarian NGOs believe that intervention to prevent people from dying or starving in large numbers is inherently good; equally, it is morally reprehensible to do nothing when people are displaced and their lives are at risk. The imperative is to action, to save lives. This categorical imperative creates the political legitimacy for action in humanitarian emergencies.

        Civil wars and complex humanitarian emergencies. Humanitarian action is occurring, however, in a context very different from the natural earthquakes and disasters that are familiar terrain to NGO personnel. Increasingly, NGOs are struggling to provide relief and assistance under conditions of civil war, often brutal civil war. In the insurgencies and counter-insurgencies characteristic of modern civil wars, human populations are the principal targets and shields. They are not the unanticipated consequences of military strategy, as they are in major conventional battles, but rather the targets of military strategies. The aim of much contemporary military strategy in civil wars is make the civilian population hostage, and, if possible, to prevent or undo the effects of emergency relief and the protection of civilians.

        In the internecine struggle for dominance in Somalia and Sierra Leone, and even more so in the openly genocidal landscapes of Rwanda and Burundi, strategies of insurgency and counter-insurgency warfare sought political control over civilian populations, inflicted costs on those populations, at times forced their movements en masse and, in some cases, systematically killed large numbers for political or military ends. Civilian casualties are not counted as “collateral damage” but as measures of strategic gain. In Somalia and Sierra Leone, militias and army units alike looted communities, destroyed available resources, engaged in scorched earth tactics against the local infrastructure, and attacked civilian populations. All over Central Africa in the 1990s, insurgency campaigns were fought behind the shields of population groups. 

        The human costs the non-governmental agencies address are not incidental to the conflict; rather, they are its essential currency. Civilians, and those humanitarian NGOs who would protect them, become the objects of military action. They and their resources stand not apart from, but directly on the battlefield. Becoming part of the battle challenges all the fundamental precepts of humanitarian action.

        Disengagement by the major powers and the consequent security vacuum. This NGO dilemma of engagement is made far more acute by the repeated unwillingness or incapacity of the major powers to act through the UN Security Council or other appropriate instruments, to provide security first for endangered civilians and then for NGO personnel who are in the field offering protection. Somalia was the exception at one stage of its emergency, but so negative were the experiences of the UN and particularly the US “military humanitarian” mission in Somalia, and so limited the strategic goals in comparison to the apparent costs, that Somalia set a “Mogadishu line” of active engagement which the US and other Western forces were thereafter unwilling to cross in the African context. The non-governmental sector found itself working in a political/security vacuum created by the decline of interest on the part of the major powers.  It is the absence of an adequate security envelope, I will argue, which creates many of the observed negative externalities of assistance and relief, and retards the prospects of conflict resolution.

        Even less demanding levels of support are dropping. The substantial increase in what the humanitarian community calls the “internally displaced” is telling: In 1991, UNHCR was responsible for 17 million refugees; by 1995, numbers had risen to 27.4 million. This increase, moreover, masks a qualitative change: the number of refugees who cross international borders and are granted asylum in another state has been declining in the last decade. The increase in UNHCR numbers are internally displaced and war-affected populations within their home countries and people outside their home countries who have not been granted asylum (UNHCR, 1995:20; Duffield, 1998: 143). The increase reflects an increasing inability for populations in distress to seek asylum across borders and become officially recognized refugees with access to the political and humanitarian rights of refugees. The growth in the numbers of internally displaced person reflects the growing tendency for the international community to disengage politically and economically from these conflicts, to attempt to contain their effects, and to ensure that the costs are internalized within the affected communities. This strategy of containment privileges relief at the expense of the protection of the basic rights of displaced populations (Duffield, 1998).

        The political economy of international assistance. As the major powers become more unwilling to engage directly or through the United Nations, they are channeling ever larger shares of their assistance to Africa through NGOs.  In 1996, more aid to Africa was channeled through NGOs than through official development assistance programs. Of course, Western government aid agencies are still the principal source of those resources, but in complex emergencies in particular, NGOs are increasingly the principal conduit of assistance and so face an ever larger share of the dilemmas humanitarianism generates in complex emergencies. The major powers expect -- unrealistically -- that the community of NGOs can fill the security vacuum left by inaction on the part of states (Lautze, Jones, and Duffield, 1998).

        In this context, NGOs have become a critical resource. For several of the worst months of the Somali famine in 1991, for example, a handful of NGOs and the ICRC were the only international presence in the country providing relief and assistance. In Sierra Leone, NGOs provided relief in parts of the country declared off limits by the UN. In Rwanda/Zaire, the flood of refugees in the autumn of 1994  was met by NGOs, working without an official UN presence. In Burundi, where military activity kept the UN out of important regions of the country, NGOs were again at the front-line in the delivery of humanitarian relief assistance.

        The root causes of the complex emergencies grow from the interlinked failures of development and the weaknesses of the state and the withering of its capacity, or the capture of the state apparatus by organized fragments of the population. In the violence that develops, social control over elements of the population is a key strategic objective of internal war, with civilians as a principal target, rather than a by-product of other military activity. Many of these internal wars fought for control over resources become cyclical and self-perpetuating, as violence generates profit for those who use it most effectively -- which often means most brutally.

        This interaction among a humanitarian ethic with an imperative to action, the withdrawal of major powers and international institutions from Africa, the ferocity of civil wars and the complexity of the humanitarian emergencies they create, and the new political economy of international assistance together generate acute contradictions for humanitarian NGOs on the front lines of conflict. Analysis of these contradictions demonstrates the fissures in the structure of international security and the challenges to conflict prevention and resolution.

The Critics:  Humanitarianism as an Obstacle to Conflict Resolution. 

        Drawing on the experience of humanitarian intervention in complex emergencies in Africa in the last several years, critics have concluded that the relief effort at best does not contribute to, and at worst, can jeopardize conflict resolution.  At least seven threads of criticism can be identified; some speak to the central dilemma of humanitarian NGOs that I have identified, while others are tangential. I begin with the most technical -- and least serious -- and progress to the most trenchant and troubling.

        1. Humanitarian NGOs are often inefficient and unprofessional. There is a significant body of critics of the operations and accountability of NGOs. The multi-donor evaluation of the Rwandan crisis could not, for example, locate a third of the 170 NGOs registered, and some $120 million of funding went unaccounted for (World Disasters Report, 1997: 12). The issue of accountability, which includes not only finance but adequate independent monitoring of performance and program evaluation, grows out of the rapid proliferation of NGOs as states began to privatize their assistance policies. The problem is serious but essentially technical. There has been reluctance by some NGOs to submit to independent evaluation, largely because independent assessments can uncover major failures (Prendergast, 1995). In the last few years, however, greater emphasis has been put on developing best practices and on monitoring of programs and performance. Especially in the context of a complex emergency, monitoring that is oriented toward support of internal evaluation and development is more likely to be acceptable and effective than external audit.

        Evaluations have also stressed the need for better coordination among NGOs, better coordination among donor governments and between donors and NGOs, more responsible and restrained use of the news media by NGOs, and stronger coordination by a lead agency within the United Nations when a complex emergency erupts (Bennett, 1996; Minear, Scott, and Weiss, 1996; Lautze, Jones, and Duffield, 1998). In Rwandan refugee camps in Zaire in 1994, it is estimated that as many as 80,000 people may have died due to poor standards of health provision (Borton, Brussett and Hallam, 1996). In Somalia, it is estimated that as many as 240,000 lives were lost due to delayed action by the international community. Furthermore, while the international response focused on food aid, perhaps 70 percent of deaths could have been averted through public health programs (Sommer, 1994: 97).

        NGOs have recognized the need for greater coordination amongst themselves and with donors and international institutions and have taken several steps to establish coordinating mechanisms. In Rwanda, for example, the NGOs established a Coordinating Committee in 1995, partly as a result of the large numbers of NGOs working in Rwanda and partly because of the tense relationship with the new government. The Coordinating Committee evaluated the broad range of  NGO programming in collaboration with the government and established an executive committee that met regularly to discuss issues of joint concern. In Burundi, the NGOs established their own forum for coordination. NGOs met regularly with heads of UN agencies to discuss joint problems and share information and, in Rwanda, were regularly represented at UN security meetings.  

        Voice. It is almost universally acknowledged that NGOs need deeper knowledge of the society, its culture, history, and language. In Somalia and Rwanda, for example, few NGOs had long-standing experience in the country, were fluent in the local language, appreciated social and cultural norms, and were experienced in working at the grass roots (Shiras, 1996). Of the large number of expatriate NGO staff in Rwanda in 1994, only a handful were conversant in Kinyarwanda. Knowledge of local parties, their networks, their purposes and strategies is critical, I will argue, to minimizing some of the negative consequences of relief assistance which fuel rather than resolve conflict. In addition to their loyalty, as expressed through humanitarian ethics, NGOs must find far better ways of giving voice to the people they wish to help.

        Finally, humanitarians need better skills in conflict resolution. In Somalia, traditional systems of authority, which did not depend on violence, and were capable of attempting the resolution of the conflict, continued to exist even after the violence erupted. A peace- building initiative sponsored by an NGO at the local level was successful because it drew upon these customary Somali conflict management practices (Menkhaus, 1997). The relief effort, in contrast, helped to cripple the traditional systems because it did not channel assistance through them but strengthened those that relied on violence (Natsios, 1997: 85-86). NGO personnel needed far greater knowledge of the local systems of conflict management and the importance of elders as authoritative voices in society.

        2. Humanitarian assistance from outside interferes with the accountability of African leaders to their populations. It reduces warring parties’ responsibility for their constituencies. This criticism is an expanded, generic version of the previous argument about technical accountability and competence. Critics allege that political accountability, through contractual arrangements, are the critical constraint on government violence against civilians and on government-induced famine. Thirty party humanitarian assistance interferes with the formation of social and political contracts within Africa.

        This criticism has been leveled most tellingly in the context of the analysis of famines. It is not natural disasters or economic collapse that create starvation and mass migration; alone, they are insufficient. Rather, famine is the result of systematic violence, deployed for political purposes, and designed to destroy the coping mechanisms and survival strategies of peasants (Sen, 1981; De Mars, 1996). Two issues arise from this analysis of the political economy of famine in a context of violence (Duffield, 1991; Keen, 1994).

        First, the argument has been made that relief assistance does not address the causes of famine, and may indeed exacerbate its severity, by making political leaders less accountable to their constituencies (de Waal, 1989). When assistance is distributed in rural areas, governments in central areas are able to avoid the political responsibility incumbent on any government, to feed their own populations (Prendergast, 1996,1997). In Sudan, for example, relief made the authorities less accountable to their civilian populations.

        Critics find it easier to diagnose the politically motivated violence of famine than to suggest strategies that can alleviate the hunger that is its consequence. They suggest that rural areas must be empowered politically so that they can forge ties with a center that becomes accountable (de Waal, 1989). Logically elegant, such a strategy ignores the context of acute insecurity created by the predatory violence that is so critical to the diagnosis.

        Analysis of the political and economic purposes of those who prey on their own civilian population does not suggest that the perpetrators are likely candidates for accountable governments. The authoritarian quality of many governments, the absence of institutions that can meaningfully hold leaders accountable, and the high levels of corruption make contractual constraints unlikely as a near-term solution to complex emergencies and violent conflict. Acknowledging these obstacles, the optimistic analysts estimate that it will take at least a decade for political contracts to form; others are even more pessimistic (Duffield, 1997). These pessimistic estimates suggest that at best, empowerment and accountability will be painfully slow processes that are unlikely to proceed in smooth, linear patterns. Political contracts cannot provide a near-term solution to violent conflict and humanitarian emergencies. Until they do, if they do, the complex emergency continues and the third party and the local humanitarian dilemmas intensify. Yet, humanitarians must contribute to the seeding of this accountability if the vulnerable populations they seek to help are ever to be given voice.

        3. Humanitarian aid is a substitute for international political action. There is a growing international indifference to humanitarian crises. Governments have privatized their assistance policies and adopted strategies of containment rather than address the underlying causes of complex humanitarian emergencies. They are increasingly resistant to accepting refugees and unwilling to grant asylum as mandated by the international refugee regime, even as they are less inclined to intervene politically or militarily to protect populations at risk. There is a corresponding decline in the public's response to the appeals of NGOs for funds. It is in this context that NGOs are being substituted for effective action by the major powers and exploited as a cover for their absence (Hendrickson, 1998).

        Humanitarian relief has also been compromised by the unsustainable and conditional consent it has accepted to access populations at risk. NGOs have experienced enormous difficulty in gaining access to vulnerable populations. These difficulties are deliberately created by warring parties who, in the context of a complex humanitarian emergency, exploit the vulnerability of civilian populations for political or military purposes. NGOs find themselves constantly renegotiating access and facing new designations of previously consented space as off limits. The warring parties in turn frequently use negotiated access agreements to build international credibility. At the extreme, this leads to the perverse outcome that the more killing is done, the more NGOs respond with additional resources. With no good choices, NGOs consent tacitly to unilateral changes in access and so empower belligerents who impose conditions that clearly violate international humanitarian law.

        4. Relief has negative consequences for development. Critics of classical humanitarian relief alleged that it had negative consequences for development, that it removed initiative and responsibility from local parties, empowered expatriates rather than community leaders, and undermined the local economy (de Waal, 1989). In response to these criticisms, some NGOs have shifted their emphasis to a new paradigm of  “developmentalist” models of relief, usually called the “relief-to-development-and-democracy (RDD) continuum.” To avoid creating a culture of dependency and to move a population toward peace as quickly as possible, relief and development should and can occur simultaneously, even while violence is ongoing (Buchanan-Smith and Maxwell, 1994).

        “Developmentalist” strategies posit a quick end to the complex emergency and a return to stability where peaceful development is possible. The fundamental elements of the strategy are local partnerships based on capacity building and the empowerment of local communities as the choosers and managers of development policies. The purpose is to create alternative livelihoods for those associated with war and a criminalized economy. The approach is multi-functional and loosely structured and, on the continuum, the boundary between relief and development blurs and, indeed, virtually disappears (Duffield, 1996; Buchanan-Smith and Maxwell, 1994).

        The concepts of local partnerships and community empowerment are key elements of a successful strategy of development, and of a process of conflict resolution that moves forward at a sustainable pace over time. Vulnerable communities must be given voice if  predators are to be constrained in any way and a sustainable process of conflict resolution is to begin. Ironically, however, the emphasis on more “participatory” emergency relief led more or less directly to the non-governmental sector’s greatest crisis of conscience and credibility. In Eastern Zaire where aid agencies were setting up camps for the influx of thousands from Rwanda, they employed the latest techniques of camp management involving, among other things “refugee self-management.” The goal is to use indigenous leadership within refugee populations to help them, as much as possible, run their own affairs. In this case, however, the leadership cadres were precisely those who had engineered the genocide and then the forced mass migration. The resulting dilemma stretched over many months, with no obvious solution.

        This new “developmentalist” model also ignores, indeed virtually wishes away, the scope of the violence and the extent of the emergency that make an early return to stability extremely unlikely. In some cases -- the Sudan, Liberia, and Somalia -- the emergency has continued for a decade or more. In other cases -- Rwanda and Sudan --  the premature declaration of an end to the emergency to fit with the new agenda is belied by the continuing, indeed, escalating violence within the country. 

        In Rwanda, the governing expectation for planning in 1996 was gradual but progressive rehabilitation and development. There were positive trends: the return of the refugees, the restoration of some basic government services, and limited economic improvement. By December 1997, however, 50% of Rwanda was again considered “insecure,” and the number of internally displaced was increasing rather than diminishing (Macrae and Bradbury, 1998). The emergency had not ended, it had ebbed briefly before intensifying again. The expectation of stability proved wildly unrealistic in the context of intensifying violence. Similarly, in Sudan, despite ongoing hostilities, an end to the emergency was declared. The government subsequently permitted NGOs to register only for development and rehabilitation, despite the growing numbers of people in desperate need of emergency relief (Lautze and Hammock, 1996: 27). The premature end to the emergency served the political purposes of a regime that was oppressing vulnerable populations.      

        The relief-to-development-to-democracy approach also creates pressure to reclassify emergencies so that the multi-functional approach can begin to work. Premature re-labeling has led to the normalizing of emergencies and the raising of thresholds of civilian violence before an emergency can be declared (Duffield, 1998). More generically, developmentalist approaches to relief seriously underestimate the difficulty of implementing development programs in the context of the acute violence and extreme insecurity that are characteristic of protracted humanitarian emergencies. They do so in part because they ignore the politics of those who benefit from the prolonged emergency.

        Finally, there is little systematic evidence to sustain the argument that relief generally displaces development and creates dependency (Carlsson, Koehlin, and Ekbom, 194:203). It may well do so under certain conditions, but we do not know enough to differentiate the conditions under which relief does block development. Given the limited amount of relief that is provided and the relatively short duration of most, though not all, large relief operations, it seems unlikely that relief would appear an attractive option in comparison to alternative coping strategies usually available to subsistence populations. It is more likely that acute violence disrupts these coping strategies and vulnerable populations have no choice but relief assistance. We need to investigate rather than assert the relationship between relief and development.

        5. Humanitarian aid emphasizes reconstruction at the expense of justice. Even when there is attention to reconstruction, it is largely focused on restoring services and rebuilding economies, not on the political accountability that is central to a reformed political system. Humanitarian relief, in part because of its commitment to impartiality and neutrality, avoids dealing with the political ambitions and past actions of predators (Duffield, 1998; Keen, 1994, 1996). This criticism of NGOs who deliver relief assistance, which is apt on its terms, applies equally, however, to the development and conflict resolution NGOs when they work in complex humanitarian emergencies. Reconstruction of any kind assumes a benign rather than a predator state or militia who systematically targets civilian populations for economic or political ends. Yet, often it is precisely those who created the massive disruption originally who are subsequently invited to participate, first in reconstruction, then in development, and, finally, in conflict resolution. All three can compromise the pursuit of justice.

        6. Humanitarian relief fuels war and conflict through asset transfer. The evidence is overwhelming that, in recent complex humanitarian emergencies, the assistance and relief that NGOs have provided to populations deliberately put at risk, have, at the same time, become the fuel for continued and renewed warfare (Duffield, 1993). In Somalia, for example, food was extraordinarily scarce as a result of drought and civil conflict and, consequently, its absolute value rose to unprecedented levels. Its high price, in the context of economic collapse, mass unemployment, and a dramatic drop in family income, increased the relative value of food. Food brought into Somalia through the relief effort was plundered by merchants, by organized gangs of young men profiteering from the black market, and by militia leaders who used the wealth the food bought to buy weapons and the loyalty of followers (Natsios, 1997). In Rwanda and Sierra Leone, as well as in Somalia, assistance has been “taxed,” or stolen to fuel processes of conflict escalation rather than promote conflict resolution.

        Resources channeled into Somalia by UN agencies and NGOs became part of a complex economy of warfare between rival militias and rival clans. Theft of those resources by militias was common. Equally significant was the ability of militias, in the absence of a security envelope for the local population and for NGO personnel, to use force and the threat of force to compel NGOs to hire some of the same forces to guard relief supplies and convoys who were the source of the humanitarian crisis (Clarke and Herbst, 1997: Prendergast, 1997). In so doing, NGOs legitimated those who were preying on local populations (Anderson, 1996). In Sierra Leone and Liberia, conflict analysts and medical NGOs learned that they could plan by following the pattern of UN food deliveries: when food was distributed to a village or displaced persons camp, the militias would quickly attack to steal the relief supplies, killing dozens of villagers as they did so.

        UN and NGO resources in eastern Zaire were subject to political control and taxation by the forces that perpetrated the Rwandan genocide of 1994. Less by theft and diversion than by controlling distribution of relief supplies and the flow of information, Rwanda’s genocidaires turned UN-managed and NGO-operated refugee camps into political and resource bases for continued and renewed genocidal warfare, both within Zaire and in western Rwanda (Duffield, 1994). When the post-1994 Rwandan regime sought to break the genocidaires’ control of the camps, civilian refugees became moving shields between two armies. Relief supplies and the NGO presence were used to lure starving refugees out of hiding in the forests of Zaire, and these refugees were then slaughtered by the tens of thousands. At the extreme, NGOs were transformed from sources of protection into resources for destruction.

        The diversion of humanitarian assets by warring parties, at the same time as they are targeting warring parties, is the most serious challenge NGOs face. It is the most dramatic example of the perversion of the humanitarian agenda and it is a serious obstacle to the resolution of conflict.  To the extent that humanitarian NGOs are inadvertently fueling the cycle of violence which is making populations vulnerable, they and those that they seek to help are trapped in a vicious process. Yet to abandon populations at risk to the predators is an almost unthinkable choice.

        Critics disagree radically on the appropriate solutions. Some urge that relief assistance be radically restructured or even eliminated. There is agreement among the radical critics that conflict can be resolved only through a long process of creating a vibrant civil society that can demand good governance, but there is considerable difference about how civil society can best be promoted. Some urge the virtual exclusion of third party humanitarians, so that governments and populations have no alternative but to create contracts, while, at the other extreme, some urge a high level of partnering between “progressive” northern and southern NGOs, to force governments to be accountable (Prendergast, 1996). Aid would be made conditional on good governance and respect for human rights.

        I have already examined the real and serious obstacles to the development of binding contractual relationships as a near-term solution to violence. In the fragmented politics of those marginalized by the global economy, some claim that even evolutionary processes toward political accountability are delusional (Duffield, 1997). Since society is fragmented, politicized, and incorporated into black or gray predatory economies, the model of a civil society separate from a centralized state does not fit; there simply is no civil society to strengthen. One pessimistic analyst of post-modern violence concludes that “War and famine do not stand out from normal social relations; they are simply a deepening of exploitative processes”(Keen, 1994:12). The same kind of contextual challenges would confront those northern NGOs who partnered with their southern counterparts in order to force local militias and predators to be accountable (Fowler, 1998).

        The larger critique -- that relief fuels war -- is valid and important, but it does not develop either criticism or solution within an appropriate context. Withdrawal of the humanitarian presence, I argue, should be only the last in a staged series of options, and even then, it has negative consequences because those that are watching and reporting to the outside world will no longer be there, even as a mild deterrent. It is also important to note that not only relief but many other economic activities fuel and sustain war as well. The importance of relief is likely to vary by context: in eastern Zaire, relief assistance was a critical resource to militia leaders, while, in other cases, the drug trade and smuggling were far more important generators of resources to predators. No study systematically investigates the proportionality of effects on war, yet only careful empirical analysis can resolve the question of the proportional impact of humanitarian aid on war. 

Minimizing the Negative Consequences of Aid in War 

        There are no easy or obvious solutions to the fundamental dilemmas humanitarian NGOs face as they seek both to help populations preyed upon by governments or militias, and to help resolve the conflict so that vulnerable populations will no longer be targets of systematic violence. Indeed, analysis of the structure and context of complex humanitarian emergencies offers little grounds for optimism about a quick end to violence. Accumulated experience in attempting to manage these emergencies and resolve the internal conflicts of the last decade is no more encouraging. For humanitarians, the dilemmas are likely to persist and intensify. We need look no further than the recent experience in Kosovo in August 1998, where civilian populations were yet again systematically targeted, humanitarians were again denied access even after consent was given, and hunger was deliberately created by the burning of crops and the destruction of farming implements.

        Two conclusions are clear from the analysis. First, complex humanitarian emergencies of the kind we have seen in this last decade in Africa are likely to continue and not only in Africa, well into the future. Second, NGOs committed to humanitarian values will continue to engage on behalf of vulnerable populations. Disengagement is not an option for humanitarian NGOs, even if it is for states. If anything, given the privatization of assistance and the retreat of the United Nations hobbled by budget deficits, NGOs will play an even larger role than they have in the past (Carnegie Commission, 1997: 105-127).The central challenge, then, from the perspective of conflict resolution, is to find ways of minimizing the negative externalities of assistance as aid flows to the most vulnerable populations. NGOs are looking for ways to prevent the transfer of assets to the warring parties, so that their work does not fuel the cycle of war. It is vital that humanitarians learn from past experience, and that they constantly evaluate their practices to assess whether alternatives exists which would minimize the negative consequences of their work in the context of a complex humanitarian emergency. In the last five years, there has been considerable progress in exploring alternative ways of reducing these negative externalities. I consider only a few of a large number of proposals and programs that have been put in place in the last several years.

        Paying explicit attention to the diversion of food aid to warring parties, NGOs have begun to distinguish types of food aid by their market value. They ask how “lootable” their assistance is.  In Somalia, for example, rice was extraordinarily attractive to looters while sorghum evoked little interest. When, for example, a food convoy organized by CARE was attacked along the Jubba River in Somalia, the thieves left without stealing when they discovered that the trucks contained sorghum (Natsios, 1997:87). Blended foods, generally less tasty, are less attractive, and foods that can be stored for extended periods of time can be hidden from predators. The ICRC, for example, moved to cooked food to reduce the interest of looters. Careful monitoring, important on its own as NGOs seek to become transparent and accountable, was remarkably successful in Rwanda and Angola in reducing diversion. Similarly, seeds can be selected so that they are less attractive to looters: those that are easily stored, that match local habits of consumption, and that displaced populations can take with them as they move to different locales are less likely to be diverted.

        In Somalia, the Office of Foreign Disaster Assistance (OFDA), a department of USAID,  tried an innovative strategy of  “monetizing” the food that was delivered. Insofar as food had become a medium of exchange, flooding the country with food would depreciate its attractiveness and diminish the incentive for looting (Natsios, 1997: 86-93). Selling cereals as well as cooking oil to merchants would permit people to buy food with their limited incomes as the price of food declined. The monetization strategy was also designed to force onto the markets all the food hoarded by organized criminals and warlords. Monetization did affect market prices by 1993 and produced enough currency to fund significant rehabilitation and reconstruction. It did not succeed, however, in reducing diversion; the drop in food prices drove the warlords to “tax” at higher levels. Only after the military intervention, did monetization accelerate and break the hold of the warlords.

        NGOs are also trying to increase the ratio of non-food to food aid within the constraints imposed by a complex emergency. There is much greater emphasis on supporting sustainable livelihoods - distribution of fishing nets where fish are available, vaccination programs against measles, a perennial killer of children in complex emergencies, and portable educational materials so that schools can continue even as populations are forced to move. None of these are easily “lootable” material that can fuel a war economy.

        NGOs have also recognized how the economic side-effects of their operations can contribute to a war economy. Collaboration among NGOs, difficult as it is, to standardize physical costs can drastically reduce the negative externalities of assistance. In Baidoa, for example, all agencies collaborated to reduce the costs of vehicles. In Rwanda, Save the Children (UK) organized some NGOs to standardize prices of housing and transport. In Goma, UNHCR and the NGOs cooperated to put a ceiling on labor costs; salaries were immediately reduced by 50% (Prendergast, 1995:20).

        Proposals have also been developed to share information, to coordinate and plan better, to improve institutional memory, and increase area expertise so that NGO personnel can learn quickly about local politics and structures. Since the genocide and mass exodus from Rwanda in 1994, some NGOs have consciously begun to develop their capacity to collect information about and analyze political and security developments that might have an important impact on diversion of aid and, more generally, on operations. MSF has an ongoing global country watch; Action Aid has created an office called Emergency Response and Information Collection (ERIC) for the Great Lakes Region; and many NGOs feed into and from the UN Department of Humanitarian Affairs’ Integrated Regional Information Network for the Great Lakes and for West Africa. UNICEF has created a global Rapid Response Team and CARE is examining how it can preposition experienced staff in areas where populations seem particularly at risk (Prendergast, 1995). NGOs recognize that they need good operational knowledge of differentiation along identity and class lines if they are to succeed in minimizing the diversion of aid to warring parties.

        In response to criticism that they have violated humanitarian space, NGOs have worked together to define more carefully the responsibilities of emergency aid and to refine the ethics of humanitarian action. International humanitarian agencies have adopted standards of performance and codes of conduct: the Code of Conduct for the International Red Cross Movement and NGOs in Disaster Relief; the elaboration of a set of technical standards in the field of water and food aid delivery by the Steering Committee for Humanitarian Response (SCHR); the development of principles and best practices for the recruitment and management of relief workers by People in Aid in the United Kingdom; and the development by the SCHR of a “claimants” charter defining beneficiary rights.

        NGOs have also worked to improved their assessment of needs within a broader model which includes black as well as official economies, an analysis of the local coping strategies of populations at risk, and an assessment of their capacities as well as their needs. A broader analytic lens helps NGOs to focus on supporting herds, or replacing implements, so that vulnerable populations can survive in the face of predators. In Somalia, for example, NGOs have begun only recently to assess local coping mechanisms and capacities. NGOs are looking at emergency assistance that simultaneously supports and sustains local community structures. Here too, they are monitoring to ascertain whether aid is reaching the intended targets. Meeting local needs and at the same time sustaining community structures and building capacity is a long term, trial-and-error process as NGO personnel learn local structures on the ground. When they can do so, the distinction between relief and development begins to blur.

A Political Humanitarianism 

        These strategies, alone or together, can reduce the scope and severity but never completely eliminate the transfer of assets to warriors and other negative externalities of aid. Analysis of these cases suggests that the more complex the conflict, the more chaotic the security markets, and the more traumatized the social order, the more important an adequate security envelope for effective delivery of humanitarian assistance (Natsios, 1997:93). For  humanitarians working in complex emergencies, acute dilemmas will occur as long as the United Nations is unable to provide security as a public good and the major powers continue to disengage and privatize assistance as a substitute for political action. There is no evidence that either trend is likely to change in the near future; on the contrary, both are likely to intensify. If they do, the range of choices for humanitarian NGOs will frequently be narrow, and, at the extreme, there will be no “good” choices to be made.

In the camps in eastern Zaire in 1994 and 1995, for example, there was considerable resource transfer, misappropriation, taxation and theft by militias. Here, the genocidaires unquestionably drew their main political support from the physical presence of the humanitarian effort; the humanitarian presence provided an economic base from which they and most important, their key strategic resource -- Rwandan civilians -- could live. The critical and agonizing issue for NGOs was whether to stay and fuel the capacity of the genocidaires to make war, or leave and abandon the civilian population that the militia had targeted and exploited. The choice was cruel and stark, a political and ethical dilemma beyond the reach of any technical solution available then or now.  NGO personnel may not be able to choose to do no harm, if by doing nothing, they abandon civilian populations at risk and violate their humanitarian ethics (Anderson,1996). In the face of those who are determined to do harm to civilians, NGOs may well be forced to choose the option that does the least harm. In an effort to reduce reliance upon militias, for example, NGOs have experimented with market-based and commercial channels in Somalia.  This approach does reduce diversion as well as the number of armed security men employed by agencies, but it can empower merchants who finance the warlords (Prendergast, 1995:9). To make the choice that does the least harm, humanitarian NGOs must situate their work in its larger political context.

Humanitarians must acknowledge and analyze the explicitly political nature of their work -- relief delivery, refugee protection, election monitoring, and conflict resolution -- in the context of a complex emergency. NGOs traditionally have argued and still argue that only strict adherence to principles of neutrality and consent of the parties can insulate relief assistance from political and military agendas (Keen and Wilson, 1994). Neutrality, it is argued, contributes to the amelioration of violence and conflict resolution by effectively inducing UN agencies and governments to provide assistance, by deterring violence by their presence on the ground and their access to the media, and by their capacity to mediate among the warring parties (Berry, 1997). I, and others, allege that the context of relief assistance has changed so radically that apolitical neutrality is no longer an option. Neutrality is appropriate in a neutral environment, but the environments of complex emergencies are generally predatory rather than neutral. If the political purposes of those who target civilian populations are ignored, NGOs will miss the inherently political nature of the relief they deliver to those targeted populations and miscalculate the politics of protecting those they seek to help.

NGOs should urge the Secretary General to provide security from private markets when public security for humanitarian operations is unavailable. This analysis suggests that the more complex the conflict, the more chaotic the security markets. Yet, the more traumatized the social order, the more important an adequate security envelop for effective delivery of humanitarian assistance (Natsios, 1997:93). Complex emergencies feed on themselves, enfeebling and even wiping away legitimate security resources, spreading chaos and violence, and generating the need for even greater security resources from outside. The cycle can only be broken if security is again supplied as a public good, ideally by the major powers acting through international institutions, or by members of regional organizations acting collectively. This analysis suggests, however, that the prospects of repairing the shredded security envelope in which humanitarian NGOs currently operate are not promising.

The major powers that are critical to authorization of a UN force are likely to consider most of the humanitarian emergencies as "discretionary" and, consequently, be unwilling to commit forces, directly or through the United Nations, to a crisis that humanitarians consider urgent. The falling budget for UN peacekeeping speaks loudly. Given the demographic and social forces that reinforce the aversion to casualties in post-industrial states, this caution can only become more pronounced over time. The "Mogadishu line" has become, at the close of the decade, a military and political firebreak that, other than in exceptional circumstances, major powers outside the region seem increasingly unwilling to cross.

The Under Secretary General for Humanitarian Affairs, Sergio Vieira de Mello, observing the general lack of willingness of members of the United Nations to provide security forces for humanitarian operations, noted that states are not at all "averse to letting humanitarian staff go where they dare not send their...invariably better equipped, better trained and better protected [troops]."  He proposed the creation of "regional humanitarian security teams" trained and equipped to support humanitarian personnel at short notice; teams would be drawn from "selected troops from a variety of nations in the region concerned" (DHA News, 1997: 5, 7-8). This proposal is consistent with the so-called "regional" or "sub-regional" approach to conflict resolution, where the responsibility for peacekeeping and security rests with the countries closest to the problem. In the wake of the terrible failure first to prevent and then to stop the genocide in Rwanda, the United States, Britain, and France supported the African Crisis Response Initiative (ACRI), a project to help train and equip a standby, rapid reaction peacekeeping force; this has yet to be put to the test. By far the most extensive trial of regional peacekeeping has been the eight-year long deployment of a multi-national force or "monitoring group" (ECOMOG) first in Liberia and more recently in Sierra Leone, by the Economic Community of West African States (ECOWAS). The record of ECOMOG has been mixed, but no more so than UN and NATO forces deployed elsewhere (Smith and Weiss, 1997; Griffen, 1999; Rowe, 1998; Scott. Minear, and Weiss, 1995).

African peacekeeping and peace-enforcing efforts have been effective in providing a security envelope where they have been deployed, but the overall pattern is nevertheless not encouraging. Forces have been infrequently deployed and the choices as to where and when to intervene have been essentially arbitrary. There is also growing concern at the United Nations about compliance with international standards in regional operations that the UN authorizes. The Secretary-General recently urged the Security Council to confirm that regional organizations have the capacity to carry out operations consistent with international norms and standards, and to put in place mechanisms to monitor regional peacekeeping forces operating under the authority of the UN (Secretary-General, 1999).

When security is scarce as a public good, the security of NGO personnel in the field is, as I have noted, not surprisingly increasingly at risk. There are, however, very limited arrangements currently in place through the United Nations to promote their security, even when they are contracted to the UN. Within the UN, the United Nations Security Coordinator (UNSECOORD) coordinates, plans, and implements safety programs and acts as the nexus for interagency cooperation on security issues, exclusive of peacekeeping forces. These arrangements are restricted to personnel engaged in operations specifically authorized by the Security Council or the General Assembly (Secretary-General, 1999: 10). In a memorandum of understanding circulated in early 1997, NGOs who are implementing partners of agencies within the UN may request to be included in UN security arrangements; to do so, they must agree to pay their share of the costs and abide by UN security guidelines. These arrangements are restricted to expatriate staff of NGOs that are implementing partners and to those employees directly engaged in fulfilling the contract; they do not include local staff, or even all expatriate staff, much less extend to vulnerable populations. It is not surprising that NGOs objected to the loss of autonomy, the inequities, and the cost. Here donors could be helpful: they could emphasize as a priority and fund security costs as part of their envelopes for humanitarian assistance and they could also press for a long overdue review of the role of UNSECOORD. Even were more inclusive agreements to be negotiated with the United Nations, they would not address the fundamental challenge of the deep insecurity of the vulnerable populations humanitarians seek to help.

When security is not being provided as a public good, as it frequently is not in a complex humanitarian emergency, NGOs should reluctantly consider urging the Secretary General to draw on private resources to provide security.  The absence of international public security forces, and the lack of effective and legitimate alternatives, empowered the militias of Somalia, Eastern Zaire, Sierra Leone and Liberia to terrible effect. It is only when security is absent that humanitarian assistance prolongs rather than mitigates violence. Under these circumstances and only under these circumstances, the UN might consider hiring paid, volunteer, professionally-trained security personnel, employed without regard to national origin and beholden to its employer rather than to any single government, to secure the deliveries of emergency assistance. The concept was seriously considered in Rwanda in late 1994. In the fall of 1994, the UN received a proposal from a British company to provide training and support to Zaire's army in order to wrest control of the camps from the militias. The idea received support from one permanent member of the Security Council, but other members rejected the idea on the basis of cost and principle.

The primary purpose of private security guards would not be to protect NGO personnel, but to avoid the need to hire local providers from among belligerents to protect convoys of relief assistance. In eastern Zaire, for example, after months of inaction, two battalions of Zairian troops were hired to maintain security in Rwandan camps under UNHCR authority. The presence of the troops significantly improved law and order in the camps and diminished the authority of the militias among the refugees (Prendergast, 1995). Even then, the Zairian troops were not impartial in the broader conflict within Rwanda nor were they mandated to deal with the central issue of separating refugees from militia leaders. At the very least, private security personnel from outside the region would not fuel the local war economy nor sustain those who prey on local populations.

This kind of proposal will not be well received within the humanitarian community and many would consider it infeasible. For both practical and normative reasons, NGOs undoubtedly would prefer to avoid such a solution. There are already indications, however, that the hiring of security guards from the private sector is acceptable under specified conditions in the humanitarian community. The ICRC prohibits the hiring of local armed escorts for relief convoys, but acknowledges that the hiring of guards to combat crime and provide security for personnel may be necessary if there is no other option. When armed guards are necessary, the ICRC recommends that they be hired from “an established security firm or the police rather than the army” (ICRC, 1995, 1997). A report recently submitted to the European Commission proposed that donors could field security units to protect humanitarian work, either from national resources or “through funding specialist third parties” (European Commission, 1999). It is worth considering whether the hiring of security guards from specialized third parties is an appropriate strategy not only to combat crime but also to mitigate the violence that flows inadvertently from current  policies. Private providers of security working under the authority of the United Nations may be the least harmful response both to the privatization of assistance and to the absence of security as a public good.

Conditionality and Exit. Finally, and only as a desperate last resort, NGOs must be prepared to consider seriously the option of withdrawal when assistance intended for humanitarian purposes is being diverted into renewed cycles of conflict. Withdrawal during an emergency flies in the face of the most fundamental humanitarian commitment and impulse to protect lives at risk: NGOs cannot justify the loss of access and witness. Yet, only if  humanitarian actors are willing to suspend delivery and withdraw presence when their assistance is forming part of a cycle of violence, can they regain sufficient leverage to retain or recapture control over delivery and management of relief supplies, and to re-convert presence into protection. When other options are exhausted, NGOs must be willing to take the necessary organizational steps to ensure that they are not part of the problems they are committed to alleviate. Strategic withdrawal can also send crucial signals to future would-be perpetrators of violence hoping to use relief resources for their own purposes.

To argue that NGOs must consider withdrawing if assets are being diverted to fuel a war economy raises operational, strategic, and ethical questions. Can NGOs withdraw in the midst of an emergency?    In the past, humanitarians have withdrawn largely when their staff were harmed or at risk -- the ICRC from Burundi and Chechnya, Caritas from Burundi -- or when necessary infrastructure was destroyed -- CARE from Mogadishu, and almost all NGOs from Liberia in 1996.

Withdrawal as a strategic choice is rare, but humanitarian NGOs have very occasionally made this choice. In eastern Zaire in November 1994, fifteen NGOs withdrew from Mugunga camp in the Goma region in the face of attempts by militias to assert political control over the camps. The decision was made in response to untenable security conditions and unacceptable ethical compromises, but also to increase pressure on the international community to respond to the security dilemma. At the same time, in a controversial decision, ECHO decided to stop all funding for NGOs serving the internally displaced camps within Rwanda, hoping to create a “push” for people to return to their homes. The impact of the withdrawal is unclear, since agencies with independent funding, that considered continued assistance as a humanitarian imperative, remained in the camps.

If humanitarian NGOs are to consider withdrawal as a strategy to influence warring parties and reluctant major powers, they need the capacity to assess the severity of the negative consequences of their aid, and a set of diagnostics that they can collectively use to judge that they may be doing more harm than good.

It is possible to identify a set of diagnostics, but with the caveat that there is significant variation within complex humanitarian emergencies and the diagnostics will be sensitive to the difference in context. The likelihood of negative externalities of assistance depends in part on the degree of coherence among militias and their capacity to organize effectively; when it is very high, as it was in Rwanda, diversion is more likely than when coherence is low, as it was in Sierra Leone. Diversion also depends in part on the popular support that militias enjoy and the political control they exercise; when it is high, as it was in Rwanda, diversion is more likely than if control is limited as it was in Sierra Leone.

The taxation of relief. The political taxation of relief is an obvious indicator that aid is being diverted. Initially, diversion can be difficult to assess since theft and hijacking can be high, but not part of a pattern of systematic political diversion. The better informed NGO personnel are about local political and military organizations, about ethnic and religious fault lines, and about local social, economic, and political structures, the more easily they will be able to distinguish simple theft from systematic diversion. Systematic political diversion, which is not reduced by the strategies we considered earlier, should trigger consideration of a coordinated withdrawal.

Failure by local authorities to cooperate in registration. A second warning light is the unwillingness of  local authorities to cooperate with the UN and NGOs to register recipients of relief assistance, especially in refugee or displaced persons’ camps, and to make lists of registrants available. The failure to cooperate in registration suggests that local authorities are seeking to supplant or subvert existing distribution mechanisms in order to divert relief assistance.  If local authorities are willing to use force to monopolize control over registration process, there is a very high likelihood that aid will subsequently become a resource for violent conflict.

Obstruction of access. Negotiation of access to populations at risk often provides predatory governments and militias with the opportunity to impose inequitable political conditions that privilege some vulnerable populations at the expense of others. Especially when access is obstructed after consent has been obtained, relief is being used as an instrument to assert control over local populations for political purposes. The government of Mobuto Sese Seko, for example, repeatedly denied access to large groups of refugees and displaced persons. Access is central to protection, support, and witness.

When NGOs recognize that their assistance is doing more harm than good, that saving lives in the short term may increase deaths over the longer term, consideration of strategic withdrawal hinges in part, but only in part, on their contractual obligations. The large NGOs, with the capacity to deliver significant amounts of assistance quickly, are almost all dependent on one of six or seven UN agencies or an agency within their home government for an implementing contract. Contracts can consequently be a constraint or an inducement to making relief conditional. UN agencies typically insist on non-negotiable rates, payment schedules and penalty clauses. Schedules and penalty clauses can work against a decision to make assistance conditional, insofar as the NGO violates the contract either by politically motivated withdrawal or by allowing waiting time for compellant strategies to work. Instead of an obstacle, agency contracts could create incentives for conditional relief. Contracts could include incentives to assist in the monitoring and reporting on abuse of vulnerable populations, and require regular reporting of agreed upon indicators of diversion of assistance. They could also reduce the penalties that are an obstacle to withdrawal, provided that withdrawal occurs within defined parameters and in accordance with agreed upon principles.

A far more important constraint on strategic withdrawal is the difficulty of collective action. A unilateral withdrawal by one NGO, no matter how large, is unlikely to be effective in constraining the behavior of predators. Even the collective withdrawal from Mugunga had only limited impact; the NGOs who withdrew continued to provide relief in other camps and the flow of resources into Mugunga continued. At least two conditions are necessary if a strategic withdrawal by NGOs is to have any impact.

First, there must be coordination among the principal NGOs who are providing assistance to act in concert. This kind of decision will not be easily reached; many NGOs continue to believe that withdrawal violates the fundamental humanitarian ethic, that it is tantamount to abandoning the most vulnerable, that it will provoke looting and violence, and that the politics of withdrawal compromise humanitarian neutrality and impartiality. The most serious criticism leveled at a strategy of political withdrawal is that it is ineffective. In the aftermath of the cessation of humanitarian aid to Rwandan refugees, violence and war increased, and several hundred thousand people died; the Great Lakes region was less violent with international humanitarian assistance than it became when that aid was withdrawn. This intense debate among humanitarians may limit the possibilities of coordination to arrangements between those who leave and those who stay, so that there can be both public statement and quiet assistance. Within the limits of the possible, consulting recipients of assistance -- rather than the predatory leadership -- as to whether agencies should remain silent or protest against abuses even if they lose their access, would empower local populations, enhance accountability, and make it easier for NGOs to reach a collective decision.

Second, a withdrawal  should be accompanied by a clearly stated set of conditions for return -- an end to diversion of relief, unobstructed access to vulnerable populations, and/or cooperation in registration of refugees or displaced persons. There are cases where conditionality has succeeded. In response to looting of cars in Eastern Equatoria, four NGOs and agencies collaborated to make continuing assistance conditional on safety on the roads, as an essential component of the larger principle of unfettered secure access. The SPLA  were concerned enough about the consequences of a cessation of aid that they made certain that the raiding of vehicles stopped. A consortium of NGOs working in southern Sudan insisted on independent access and monitoring as conditions of continued assistance. Only if withdrawal is coordinated and strategic, if the conditions NGOs set can be met by the targets, can concerted withdrawal have any impact whatsoever on the behavior of a predatory government or militia. 

In the Eye of the Storm 

        Humanitarian NGOs have become important participants as assistance has been privatized and great powers interests and commitments have waned. They remain loyal and committed to humanitarian ethics, to the promotion of the welfare of those most at risk and most vulnerable. To do their work effectively, NGOs now recognize that those they seek to help must have voice, and it is their voice that, wherever possible, must be heard and taken most seriously.

        Finally, in large part because of the failure of the wider international community to provide security as a public good, humanitarians increasingly find themselves in a cruel dilemma. In complex humanitarian emergencies, where security is absent, some of the assistance NGOs provide has gone to those who prey on the vulnerable and has fueled the cycle of violence. Far from contributing to conflict resolution, they have inadvertently contributed to conflict escalation.

        The political strategies that I have outlined, alone or in combination, can alleviate some of these negative consequences, under some circumstances. They are, however, no panacea. At the extreme, the humanitarian imperative compels exit, not presence. As controversial, and as unwelcome as these recommendations are, they must be taken seriously if humanitarian space is to be preserved. Humanitarians must consider the politics of their presence in complex emergencies seriously and, in so doing, will inevitably come to consider the politics of their absence.



The project was funded by the United States’ Institute of Peace, CARE Canada, the Department of Foreign Affairs and International Trade (Ottawa), the Department of National Defence (Ottawa), and the Walter and Duncan Gordon Charitable Foundation.


Anderson, Mary. 1996. Do No Harm: Supporting Local Capacities for Peace Through Aid. Cambridge MA: Local Capacities for Peace Project, The Collaborative for Development Action.

Bennett, Jon. 1996. Coordination, Control, and Competition: NGOs on the Front Line. In James Whitman and David Pocock, eds. After Rwanda: The Coordination of United Nations’ Humanitarian Assistance. New York: St. Martin’s Press.

Berry, Nicholas. 1997. War and the Red Cross: the Unspoken Mission. New York: St. Martin’s Press.

Borton, John, Emery Brusset, and Alistair Hallam. 1996. Humanitarian Aid and its Effects. Joint Evaluation of Emergency Assistance in Rwanda. Copenhagen: Danish International Development Agency (DANIDA).

Buchanan-Smith, M. and Maxwell, S. 1994. Linking relief and development: an introduction and overview. Institute of Development Studies Bulletin 25: 2-16.

Carlsson, Jerker, Gunnar Koehlin, and Anders Ekbom. 1994. The Political Economy of Evaluation: International Aid Agencies and the Effectiveness of Aid. London: Macmillan Press.

Carnegie Commission. 1997. Preventing Deadly Conflict, Final Report of the Carnegie Commission on Preventing Deadly Conflict. New York: Carnegie Corporation of New York.

Clarke, Walter and Jeffrey Herbst, eds. 1997. Learning from Somalia. The Lessons of Armed Humanitarian Intervention. Boulder, Colorado: Westview Press.

Department of Humanitarian Affairs, 1997. DHA News, 23 (New York: United Nations).

de Waal, Alexander. 1989. Famine that Kills: Darfur, Sudan, 1984-1985. Oxford: Clarendon Press.

Delmas, Phillippe. 1997. The Rosy Future of War. New York: Free Press.

DeMars, William. 1996. Mercy without Illusion: Humanitarian Action in Conflict. Mershon International Studies Review 40, 1: 81-89.

Duffield, Mark. 1991. War and Famine in Africa. Oxford: Oxfam Research Paper 5, Oxfam Publications.

Duffield, Mark. 1993. NGOs, Disaster Relief, and Asset Transfer in the Horn: Political Survival in a Permanent Emergency. Development and Change 24:  131-157.

Duffield, Mark. 1994. The Political Economy of Internal War: Asset Transfer, Complex Emergencies, and International Aid. In Anthony Zvi, ed. War and Hunger: Rethinking International Responses to Complex Emergencies. London: Zed Books.

Duffield, Mark. 1996. Symphony of the damned: racial discourse, complex political emergencies, and humanitarian aid. Disasters 20, 3:173-193.

Duffield, Mark. 1997. Post-Modern Conflict, Aid Policy, and Humanitarian Conditionality. London: Department for International Development (DFID), Emergency Aid Department.

Duffield, Mark. 1998. NGO Relief in War Zones: Toward an Analysis of the New Aid Paradigm. In Thomas G. Weiss, ed. Beyond UN Subcontracting: Task Sharing with Regional Security Arrangements and Service-Providing NGOS. New York: St. Martin’s Press. 139-159.

European Commission. 1999. Security of Relief Workers and Humanitarian Space. Background Document to the Commission Working Paper. Brussels: European Commission, ECHO: April.

Fowler, Alan F. 1998. Authentic NGO Partnerships in the New Policy Agenda for International Aid: Dead End or Light Ahead? Development and Change 29: 137-59.

Gardener, Leon and Thomas G. Weiss. 1996. Pluralizing global governance: analytical approaches and dimensions. In Gordenker and Weiss, eds. NGOs, the UN, and Global Governance. Boulder, Colo: Lynne Rienner: 17-47.

Griffen, Michele, 1999. Retrenchment, Reform and Regionalization: Trends in UN Peace Support Measures. International Peacekeeping (Summer): 21-25.

Hendrickson, Dylan. 1998. Humanitarian action in protracted crises: the new relief ‘agenda’ and its limits. Network Paper 25 .London: RRN.

International Committee of the Red Cross (ICRC). 1995. Working Paper on Armed Protection of Humanitarian Assistance. Geneva: Council of Delegates.

International Committee of the Red Cross, 1997. Seminar on the Security of Humanitarian Personnel in the Field for Non-Governmental Organizations. Geneva: Graduate Institute for International Studies, 5 December.

Jones, Bruce and Janice Gross Stein. 1999. “NGOs and Early Warning: the Case of Rwanda,” Synergies in Early Warning. New York: CIAO, Columbia University Press.

Jones, Bruce, Janice Gross Stein, and Michael Bryans. 1999. “Doing Good Work Amid Bad Options,” Coming to Terms 3.Toronto: Program on Negotiation and Conflict Management.

Keen, David. 1994. The Benefits of Famine: A Political Economy of Famine and Relief in Southwestern Sudan, 1983-1989. Princeton: Princeton University Press.

Keen, David. 1996. Organized Chaos: Not the New World We Ordered. The World Today 52, 1:14-17.

Keen, David and Ken Wilson. 1994. Engaging with Violence: A Reassessment of Relief in Wartime. In J. Macrae and A. Zwi, eds. War and Hunger: Rethinking International Responses to Complex Emergencies. London and New Jersey: Zed Press.

Kouchner, Bernard. 1993. Le malheur des autres (The Misfortunes of Others). Paris: Odile Jacob.

Lautze, Sue, Bruce D. Jones and Marc Duffield. 1998. Strategic Humanitarian Coordination in the Great Lakes Region 1996-1997. New York: Office for the Coordination of Humanitarian Affairs.

Lautze, Sue and John Hammock. 1996. Coping with Crisis, Coping with Aid: Capacity Building, Coping Mechanisms and Dependency, Linking Relief and Development. New York: UNDHA.

Macrae, Joanna and Mark Bradbury. 1998. Aid in the Twilight Zone: A Critical Analysis of Humanitarian-Development Aid Linkages in Situations of Chronic Instability. London and Providence, Rhode Island: A Report for UNICEF, ODI/Humanitarianism and War Project.

Menkhaus, Kenneth. 1997. International Peacebuilding and the Dynamics of Local and National Reconciliation in Somalia. In Walter Clarke and Jeffrey Herbst, eds. Learning from Somalia: The Lessons of Armed Humanitarian Intervention. Boulder: Westview Press: 42-63.

Minear, Larry, Colin Scott, and Thomas Weiss. 1996. The News Media, Civil War, and Humanitarian Action. Boulder: Lynne Rienner Publishers.

Natsios, Andrew S. 1997. Humanitarian Relief Intervention in Somalia: The Economics of Chaos. In Walter Clarke and Jeffrey Herbst, Learning from Somalia: The Lessons of Armed Humanitarian Intervention. Boulder: Westview Press: 77-95.

Prendergast, John. 1995. Minimizing Negative Externalities of Aid: the Ten Commandments. Paper prepared for the United States Institute of Peace.

Prendergast, John, 1996. Frontline Diplomacy: Humanitarian Aid and Conflict in Africa. London: Lynne Rienner.

Prendergast, John. 1997. Crisis Response: Humanitarian Band-Aids in Sudan and Somalia. London: Pluto Press.

Rowe, Sylvester Ekundayo. 1998. ECOMOG C  A Model for African Peace-Keeping. African Law Today.

Scott, Colin, with Larry Minear and Thomas G. Weiss, 1995. Humanitarian Action and Security in Liberia, Occasional Paper #20 (Providence, RI: Thomas J. Watson Jr. Institute for International Studies, 1995)

Secretary-General of the United Nations. 1999. Report of the Secretary-General to the Security Council on the Protection of Civilians in Armed Conflict. New York: United Nations. IVG 34.    

Sen, Amartya. 1981. Poverty and Famines: An Essay on Entitlement and Deprivation. Oxford: Clarendon Press.

Shiras, Peter. 1996. Humanitarian Emergencies and the Role of NGOs. In James Whitman and David Pocock, eds. After Rwanda: The Coordination of United Nations Humanitarian Assistance. New York: St. Martin’s Press.

Smith, Edwin M. and Thomas G. Weiss. 1997. UN Task-Sharing: Towards or Away from Global Governance. Third World Quarterly.

Sommer, J.G. 1994. Hope Restored? Humanitarian Aid in Somalia 1990-1994. Washington: RPG.

Smith, Steven R. and Michael Lipsky. 1993. Non-Profits for Hire: The Welfare State in the Age of Contracting. Cambridge, Mass.: Harvard University Press.

UNHCR, 1995. The State of the World’s Refugees 1995: In Search of Solutions. Oxford: Oxford University Press for the United Nations High Commission for Refugees.

Voutira, E and S.A. Whishaw Brown. 1995. Conflict Resolution: A Review of Some Non-Governmental Practices -- A Cautionary Tale. Oxford: Refugee Studies Programme, Queen Elizabeth House.

World Disasters Report. 1997. IFRC/RCS. 

Zartman, I William. 1995. Collapsed States: The Disintegration and Restoration of Legitimate Authority. Boulder: Lynne Rienner.