Journal of Humanitarian Assistance
As RPF units rapidly advanced to seize power in mid-July, members of the old regime led a massive flow of civilian refugees across the border to Tanzania and Zaire. Among them were some 30,000 government soldiers, militia members, local officials and former national leaders. In the Bukavu area in South Zaire, government soldiers formed separate camps from the very beginning; in the Goma area, the army, the militia and the civilian refugees were more thoroughly mixed. In Tanzania, the military were not as visible among the refugees, but the militia and former officials were.
These refugee settlements were problematic. There was little security in the camps. Food and other relief supplies were diverted from the intended beneficiaries. The old leadership used its hold on the camps to intimidate refugees and prevent repatriation. The areas became a sanctuary allowing the defeated regime to regroup, rearm, and launch incursions across the border. With ex-FAR officers openly proclaiming their intention to "kill all Tutsi who prevent us from returning", the camps became a key element in a new cycle of conflict.[120]
While "refugee-warrior communities" of this kind are characteristic of several exile movements, the magnitude of the Rwandese refugee population, and the issues of guilt arising from the apparent widespread participation in the genocide, made the situation particularly difficult with respect to both relief and conflict management. The search for solutions was also precedent-setting in many ways. For the first time, the UN Security Council discussed security in refugee camps as a peacekeeping matter. While the Council did not come to a conclusion, the eventual solution devised was innovative and represented a measure of success.
6.1. The refugee camps: humanitarian and security issues
It was soon painfully obvious that the perpetrators of human rights abuses and genocide were fed and assisted in the camps. According to the Statutes of UNHCR and subsequent UN authorizations, as well as the OAU Refugee Convention, all refugees from war are prima facie entitled to protection, but the exclusion clause of the 1951 Refugee Convention applies if there is reason to believe that a person has committed serious war crimes or crimes against humanity. The legal norms, in a word, are unambiguous. To apply them in the Rwanda case was another matter. Registration is necessary to determine eligibility, but the rapid influx and vast numbers involved made it impossible to register the refugees upon arrival. Even when certain individuals were known prima facie to be excludable, they could not easily be removed. International relief organizations and UNHCR encountered the problem as soon as the first refugees arrived in the Ngara region of Tanzania.
The rapid influx into Tanzania of some 250,0000 persons during a couple of days in April created mass confusion. However, former provincial and commune officials soon reconstituted themselves in the camps, thereby gaining considerable control over the food distribution as well as the refugees in Ngara.[121] Among them was a former bourgemestre and known killer. Attempts by UNHCR to remove him from the camp resulted in a riot including several thousand people. The incident _affected aid workers and refugees alike. Some NGOs, including MSF (Médecins sans frontières) suggested that the only alternative was to leave the camps. This option was actually carried out by MSF, which suspended most medical aid and some water services. UNHCR and most NGOs decided to stay in recognition of the humanitarian imperative to protect and assist a vast population of refugees, even if that meant assisting persons guilty of crimes against humanity.
The "Gatete incident" - named after the notorious bourgemestre - accentuated the immediate need to police the camps. While policing would not address the problem of removing the excludables, it would deprive them of some power and also improve security for the refugees. An agreement was reached between UNHCR and the Tanzanian government that recognized the legal fact that the formal authority and ultimate responsibility for policing refugee populations lies with the host state. Tanzanian policemen, paid by the government but supported in other ways by UNHCR, policed the camps and arrested sufficient numbers to establish the principle that violations could no longer be carried out with impunity.
The problems in Goma were far more intractable. There, the refugee inflow was much larger - 1.2 million refugees fled into Zaire in a matter of days - and the soldiers and militia were more numerous. The refugees established themselves in a corner of Zaire remote from central authority and possibly any system of accountable governance. Soldiers and militia carried their weapons with them; those who were disarmed by Zairian troops at the border were usually able to recover or replace them with the assistance of other Zairian forces. Altogether, these factors led to an initial condition of chaos and anarchic violence; later, it helped the former leaders to transform the refugee community into a state-within-the-state.
Armed, organized groups of ex-FAR soldiers and militias were part of the sprawling refugee settlements in Goma. In late August, after the devastating cholera epidemic had passed and the camps were taking shape, a flare-up of violence occurred. The absence of any accountable authority in the camps was one factor. The effort by the former leadership and armed elements to assert their power over the refugees was another. The purpose was twofold: to prevent the refugees from returning, legitimizing in that sense the new government in Kigali, and to secure a popular base for the former regime in its continuing struggle for power with the RPF. The refugees thus became hostage to the ambitions of the former leaders, as well as victims of ordinary crime. Killings, threats, extortions, rape and other physical abuses were common in the Goma camps throughout the autumn. Relief distribution was controlled by the military and the former leaders who used the traditional Rwandese administrative community unit of cellule to divert food. Refugees who tried to return home were intimidated or killed. While conditions within Rwanda did not encourage return, violent harassment in the camps helped to bring spontaneous repatriation to a virtual halt after the first wave of return ended in early August 1994. UNHCR attempts to start registration in September-October had to be aborted when its personnel were threatened. Threats to relief officials continued in the autumn, culminating in November when 15 international relief organizations informed UNHCR that they would be forced to leave the camp if the security situation did not improve.
Most of the retreating FAR units in the Goma area had established themselves with their families in a separate camp (Mugunga). Later, other soldiers consolidated their presence in a second military camp. This self-induced separation eased the humanitarian dilemmas for UNHCR and organizations whose function was to assist only civilian refugees or military hors de combat. The military could still move in and out of the refugee camps and assert their influence, but the more immediate problem was the militia.They were present throughout the camps, believed to be behind much of the violence, and - to the extent that they had played a major role in the genocide - were not even entitled to protection and assistance as refugees. Local Zairian troops who were called in to quell riots had proved to be part of the problem rather than the solution.
6.2. Searching for solutions
As the lead agency in refugee matters, UNHCR early on raised the issue of camp security with the Zairian government. In September 1994, a joint mission from UNHCR and the government of Zaire considered the feasibility of separating the militia from the rest of the population. The mission estimated that around 100,000 persons, consisting of militia members and their families, would have to be moved. The costs and problems of identifying, separating and relocating this group - by force if necessary - were considered prohibitive. UNHCR fell back on a more modest proposal to police the camp with a security contingent drawn from Zaire's elite forces, backed by international technical expertise.
The High Commissioner for Refugees, Sadako Ogata, obtained agreement from the relevant Zairian authorities and forwarded the proposal to Boutros Boutros-Ghali at the end of September. The proposal was almost identical in format to the later Zairian Camp Security Operation (ZCSO), which finally was agreed upon. However, the agreement came only after the UN Secretariat and the Security Council had debated and deliberated for over three months, rejecting the initial proposal and examining counter proposals in the interval. In the end, UN/New York asked UNHCR to unearth the original draft (S/1995/65). Meanwhile, security conditions in the Zairian camps worsened - in one incident in Goma in late November, 21 persons were killed and 77 wounded; intimidation of refugees wanting to repatriate continued; and the former leadership strengthened its hold.
The initial reaction of the Secretary-General to the UNHCR proposal was negative. In part, Boutros-Ghali was disinclined to rely on Zairian forces due to the political sensitivities of working with President Mobutu and, in effect, helping to finance his elite troops. More fundamentally, the UN Secretariat viewed the question of security in camps as more than a protection issue and rather a matter of peacekeeping since it involved separating civilians from armed groups and, in principle, dealing with the latter as a threat to both the refugees and the security of Rwanda. So defined, the matter clearly went beyond the mandate and competence of UNHCR and required the involvement of the Security Council. The Secretary-General consequently took charge of the issue and instructed DPKO to prepare options for a comprehensive solution.
While security problems in refugee camps were common and recurring issues, the Security Council had not previously considered them to be a matter of peacekeeping. UNHCR officials remained sceptical to this interpretation, citing tension between humanitarian concerns and military operations. When General Dallaire floated a proposal to extend UNAMIR II's operations to the camps in Zaire, UNHCR opposed it on the grounds that UNAMIR had shown insufficient understanding of the humanitarian aspects of the emergency and of voluntary repatriation. Some countries that had contributed units to UNAMIR were also reluctant to extend the mandate to operations across the border. As one official said, it seemed risky to "send our boys into the black pits of the camps in Zaire." A similar sentiment prevailed when the Security Council discussed the options laid out in DPKO's report of 18 November (S/1994/1308).
The Secretariat's maximum option was to make a comprehensive assault on the problem by not only separating soldiers and militiamen from civilians, but also disarming the first two (Jones 1995a). The operation was seen to require 7,000 well-armed troops, with 4,000 for the Goma region and 3,000 for the Bukavu area, and a Chapter VII enforcement mandate. Having hesitated to send fewer than that to Rwanda in UNAMIR I or II, the Security Council dismissed the option as "fantasy", as one participant later said. Option B was a slimmed-down version of A, involving 3-5,000 troops and a Chapter VI mandate. It was received with more interest, but when members were asked to contribute troops, only one out of 60 states approached responded in the affirmative.
Simultaneous on-site investigation by a joint DPKO-UNHCR team concluded that the task, at any rate, was more difficult than anticipated and would require more than 5,000 troops. That left the third option of privatizing the security service. A British security company offered to train and provide logistical support to Zairian troops. While attractive to some Security Council members, who saw it as a way to depoliticize a sensitive issue, others argued that it amounted to shirking an international public responsibility. UNHCR, which would have to raise the funds, judged it much too expensive.
The Security Council failed to support any of the options tabled. In early January, Boutros-Ghali acknowledged that his efforts to find a suitable plan for dealing with the situation had failed and asked UNHCR to revive its original proposal of September 1994. This materialized in the form of the Zairian Camp Security Operation (ZCSO).
6.2.2. The Zairian camp security operation
In a novel development, UNHCR and the government of Zaire signed a memorandum of understanding on 27 January 1995, whereby the latter agreed to provide a contingent of elite troops to police the refugee camps, working with an international civilian police unit. UNHCR would pay their salary, food, health care and other incentives. The operation was carefully designed to meet demands from the Zairian government that its troops would serve only under a national command structure.122 The Zairian troops would be trained and liaise with an international group of civilians with police and military backgrounds, but would remain under the command and control of their own officers. The first group of 100 men from the prestigious Presidential Guard arrived on 11 February; by the end of April the contingent was increased to a total of 1,513.123
By mid-1995, the operation as narrowly defined had successfully established security in the camps. The Zairians had been trained in refugee law and riot control, and worked closely with the small international civilian police (CLSG). Their task was to improve law and order conditions in the camps; prevent intimidation and violence against refugees wishing to repatriate; protect relief personnel, infrastructure, equipment and supplies; and escort returnees to the border. Reports from both UNHCR and other sources registered progress towards these ends. The numbers of killings, theft, banditry, rapes, beatings, and other safety incidents had been greatly reduced, and the associated tensions in the camps had diminished. Relief workers found it easier to perform their duties. The critical impunity question had been addressed as people were arrested and punished for criminal acts. Small arms were seized, although systematic searches were not conducted or envisaged.[124] Ex-government officers who made political speeches in the refugee camps were investigated and in some cases arrested (Halvorsen 1995). The effect on repatriation was more uncertain. UNHCR figures for repatriation show a steady decline during the first three months after the ZCSO was established, but, after a low point due to the Kibeho massacre in April 1994, a steady climb was recorded from May to July (See Study IV).
Given the importance of camp security issues in refugee situations worldwide, it is important to consider the factors behind this relative success story. The most important of these appeared to be the elite nature of the Zairian troops, their careful training and substantial remuneration, and their close liaison with the international civilian contingent.
Yet the program did not deal with the security issues in a broader sense. The former government soldiers were still in their separate camps; the militiamen and the military had easy access to the regular refugee camps from which they recruited and trained new soldiers. Their equipment was replenished with new arms supplies. Their ability to attack across the border and, in the longer run, to mount a serious security threat to the border area and the new government of Rwanda was unimpaired, indeed it strengthened over time.
The UNHCR solution, in other words, was based on a fragmentation of the problem rather than a comprehensive assault as envisaged in the alternative peacekeeping approach. Separated out, the military component of the security issue was left unattended. UNHCR had neither the mandate nor the competence to take on a rag-tag army in exile. The Security Council had declined to act, although the camps were clearly a matter of international peace and security. The OAU was silent, even though the conflicts wrought by refugee-soldiers permeate the history of post-independence Africa, including Rwanda's. Zaire, itself in a condition considered precarious by many, made no efforts to accept its minimal legal responsibility to prevent the exiled armed groups from making incursions across the border.
While some of the troops retreating into North Kivu were disarmed by Zairian troops at the border, many weapons were either stocked for later use or replaced by new ones. Virtually all levels of Zairian authority were involved in channeling arms to the ousted Rwandese government troops, including national and provincial authorities, the armed forces (FAZ) and semi-private cargo companies.[125] As documented by human rights organizations, supplies were delivered across the border via Goma airport during the war (April-mid-July), and continued to be supplied afterwards (late July and August 1994). Zairian authorities helped ex-FAR units and staff to establish themselves in five sites along the border in the North and South Kivu areas. Thus, with Zaire's help, the defeated forces could regroup, retrain and rearm freely. The Zairian government in this respect violated a number of international legal instruments - the UN arms embargo on Rwanda ((S/RES/918(1994)), the 1969 OAU Convention that proscribes armed encampments of refugee-soldiers close to the border, and international law prohibiting armed incursions across the border - as well as the legal and moral principles to punish perpetrators of genocide.
Some NGOs and Western journalists reported that France violated the UN arms embargo by being involved in arms deliveries to the retreating FAR units via Zaire in May-July 1994 (see Study 1). Informed French military officials acknowledge that there undoubtedly was a clandestine arms flow across the Zaire-Rwanda border in this period, but deny that France was involved. The French Ministry of Foreign Affairs also has categorically denied all reports of French involvement.[126] None of the allegations are supported by documentary evidence - which at any rate would be hard to obtain given the nature of the transaction - but rely on unnamed local witnesses, or statements by named individuals that later were retracted. One crucial piece of documentary evidence points in the opposite direction. About one month after the genocide commenced, a delegation from the Rwanda interim government met with the head of the Military Mission in the Ministry of Cooperation in Paris, General Huchon, and pleaded for political and military support. According to the report the delegation sent back to Kigali, Huchon told them in no uncertain terms that the French military was prevented by public opinion and the media from giving any support; "their arms and legs were tied".[127] On the other hand, it is clear that the French army and the military intelligence agency possessed the structural prerequisites for providing clandestine aid, and that four years of close collaboration with the Rwandese army would have given plausible motives for assisting an erstwhile client in distress.
The evidence implicating other countries in violation of the spirit, if not the letter, of the arms embargo is less compelling, although actual or prospective supplies from South Africa and China have been reported. Since the retreating government forces brought the reserves of the Central Bank along with them, they had easy entry to a large and poorly controlled international market of dealers in small arms. A UN expert commission was appointed by the Security Council in September 1995 to examine these and other reports of violations of the arms embargo. Its first report was expected in late January 1996.
In addition to a large external refugee population, war and genocide had created a large number of internally displaced persons, especially in the French "safe zone" where an estimated 1.2-1.5 million people had fled towards the end of the war. For the new Rwandese government, the concentrations of internally displaced represented an internalized version of the problems posed by the camps across the border.
6.3. Internally displaced: Kibeho [128]
In the aftermath of Opération Turquoise, up to 390,000129 internally displaced persons (IDPs) were left in 33 camps in the south-west of Rwanda. In the autumn of 1994, the government decided to close the camps by December 1994, using force if necessary.[130] The NGOs, ICRC and UN agencies[131] insisted on voluntary return. In November, with the government, they developed an Integrated Humanitarian Response. Opération Retour was launched in December of 1994. But it ground to a standstill in February 1995 because the IDPs were no longer willing to return to their homes voluntarily because of the deteriorating situation in the home communes and the intimidating actions of the hard core within the camps.[132] An Integrated Operations Center (IOC) working group was established by the government, UN agencies and NGOs on 6 February 1995, and an IOC was set up within the Ministry of Rehabilitation and Social Integration (MINIREISO) in March of 1995.[133] Increasing insurgency across the borders heightened the security concerns of the government. RPA shootings, beatings, and harassment in camps had increased in February.[134] By the end of February, the IOC concluded that Kibeho camp, with almost half of the remaining IDPs, was becoming a criminal sanctuary and possibly a center for recruitment and training of insurgents.
The government once again determined to close the camps, using force if necessary. In early March, the Integrated Task Force developed six options[135] to reconcile the legitimate aspirations of the Rwandese state, on the one hand, to establish stability on its territory and prevent genocidal killers who sheltered in the camps from escaping justice, and, on the other hand, the concern of the UN and the NGOs with protecting the rights of individuals and avoiding exposing innocents to unnecessary risks. The first option focused on independent forceful action by the RPA; it predicted that a mass chaotic dispersion would result, with large numbers heading across the border to Burundi, and large numbers of deaths, particularly children. Such action would also violate the agreement of cooperation between the international agencies and the government, instill distrust of the government in the people, and create a rift between the international community and the government. On the other hand, continuing with a plan of voluntary return would violate government policy and was not likely to be effective, but was likely to consolidate extremist control of the camps. The UN, NGOs and government agencies were faced with a real dilemma.
The first option was rejected. Four other options focused on various ways to expedite a voluntary program, and they too were dismissed because, among many reasons, they did not satisfy government policy and would lead to consolidation of extremist control over the camps. The other option dealt with avoiding the use of force while also surrendering the principle of strict voluntary return. A variation of this option was adopted.
The decision was made to close the camps in series by gradually reducing food supplies and transferring the displaced to their home areas. Those who chose to remain in the camps after the returnees left would be investigated as perpetrators of the genocide. The plan would be accompanied by a program of information in both the camps and the home communes, strengthened security and administration of justice in both places, and an escort service jointly conducted by the RPA and UNAMIR to guarantee safe conduct, complemented by food stations manned by NGOs along the way. The operation would begin about 6 April.
The commencement date was postponed until mid-April, then 18 April, and then there was an agreement to postpone again until the 24th, but there is confusion on whether the latter date was communicated to the RPA.[136] When the cooperative plan for vacating the camps seemed to break down, there were late warnings that a number of casualties could be expected if the plan was executed improperly.[137] Given these late warnings, NGOs, though part of the IOC, subsequently questioned why the UN had not sent in more troops. For some NGOs, the UN should have known that massacres would occur given the large build-up of troops. However, as a UN official pointed out, the fact of a troop build-up does not necessarily mean that the soldiers will be used in precipitous action against the IDPs, and is not sufficient grounds for predicting a massacre.
On 18 April, the RPA surrounded Kibeho camp with two battalions and cut off the food supply. The Minister said that the humanitarian community had agreed to that date.[138] A population of 80,000 was squeezed by the cordon from five hills onto one hill.[139] On 18 April, 8-11 children were killed in a stampede. On 19-20 April, 13-22 IDPs were killed by the RPA opening fire when IDPs apparently threw stones or started to snatch weapons. On 20 April in the late afternoon it started to rain. A large group of IDPs, deprived of food, shelter and sanitation for three days, were either stampeded or panicked and tried to break through the cordon. The RPA allegedly suffered casualties from firing from IDPs.[140] The RPA fired into the crowd; numerous IDP casualties resulted. The cordon was restored. In late afternoon, when it was raining hard, the cordon was breached again. There was more firing, more RPA casualties, more return fire by the RPA with numerous IDP casualties. The RPA even used rocket-propelled grenades. On the night of 22 April, there was both sniper fire and machete attacks among IDPs.[141] The evacuation of the camp had deteriorated into a full-scale battle using innocent victims as expendable tools of war. On the fifth day, the largest massacres occurred.
The estimated number of persons killed between 18 and 23 April ranged from a low of 300, given by the Rwandese government, and a high of 8,000 according to Australian UNAMIR soldiers who made a partial count. Official UN estimates were revised downward soon after 23 April to 4,000 and then 2,000. The Special Representative of the Secretary-General, Sharyar Khan, and the UNAMIR Force Commander held that the figure was around 2,000. The deaths were caused by gunfire, machetes and trampling; the International Commission of Inquiry found that many of the dead had machete wounds. Since the RPA did not have machetes, this suggests that the hard core elements in the camps were responsible. However, there was also evidence of summary executions of IDPs by the RPA.
Though there was a serious problem in the implementation of the agreed plan,[142] and specific confusions over the starting date for the operation, there were deeper problems, specifically the different priorities, perspectives and values of the various parties, and the impact of the genocide on the perceptions the parties had of each other. The repeated shift in commencement dates consolidated RPF suspicions about UN and NGO intentions. The UN had failed to comprehend adequately the urgent security concerns of the government.
What or who was responsible for this appalling loss of life? There were many possible candidates. The terrible circumstances played a key role: the internally displaced were kept in camps; Opération Turquoise had failed to arrest the militants; the alleged perpetrators of the genocide continued to use their own Hutu compatriots as shields and as potential cannon fodder to launch attacks on the government and even used machetes to enforce obedience within the camp.
The underequipped and impoverished government, still threatened by the militant elements from the defeated genocidal regime, could easily be faulted. There was a clear breakdown in communications and authority among various governmental sectors.[143] The International Commission was clearly correct in faulting the RPA for its lack of communications, its inexperience, its inappropriate equipment and training for what was essentially a police action. But these were not the essential elements; none of them explain the RPA's precipitous action,[144] the most immediate cause of the disaster.
Could the RPA have been stopped? UNAMIR II had a mandate to support and provide safe conditions for displaced persons. The mandate included taking self-defense actions against those who threatened populations in protected sites and the distribution of humanitarian relief. But the humanitarian agencies and UNAMIR II had agreed to the suspension of humanitarian relief, albeit gradually rather than suddenly. Moreover, several UNAMIR units were at the time on joint patrols to escort the bulk of the population to their home communes. Should the Force Commander have abandoned the returnees, immediately consolidated his contingents from all over Rwanda, and ordered the superior forces of the RPA to cease and desist? UNAMIR II was also tinged with the critism from the RPF that the previous UN force had created only an illusion of security prior to 6 April, and had been withdrawn when the genocide commenced. If confronting the RPF at Kibeho, the UN force would be subjected to the accusation of protecting the killers rather than trying to arrest them. For these reasons, UNAMIR was impotent to do anything.
The NGOs understandably placed primary emphasis on the well-being of their humanitarian charges, perhaps without giving adequate consideration to the predicament and need for both security and justice, and, most importantly, the sense of urgency of the government. The International Commission criticizes the NGOs for allegedly encouraging IDPs to remain in Kibeho, but cites no evidence for this charge. Government sources close to the event, and whose primary interests differed from those of the NGOs, found little to substantiate the Commission's conclusion.145 However, tension between the command and control structure of UNAMIR and the anarchic system of NGO coordination did not help the efficacy of the return operation.
A more serious factor was the failure of the international community to invest adequate resources in Opération Retour to initiate the information campaign in both the camps and the home areas as well as rehabilitate the justice system and the home communes in preparation for the returnees, particularly in the winter phase of the plan when the operation was still a voluntary one. There was also a failure to act sooner and use the not-so-gentle persuasion of promising to withhold food when the incentives to return home to take advantage of the January crop season were stronger, before the extremists could organize the resistance, and when the IDPs were not subject to the dismal feedback by returnees and new IDPs about the perilous conditions at home.
What was not at fault was the effort to achieve coordination and the willingness to devise a plan that attempted to be both humane and realistic. The UN and the NGOs, however, were not realistic enough in their lack of a sense of urgency and reluctance to employ not-so-gentle methods earlier, while the RPA clearly lacked sufficient humanitarian sensibility for the well-being of the innocents, particularly women and children. In the end, a clear division of responsibilities and explicit lines of authority in the implementation of a coherently drawn-up camp evacuation might have averted the tragedy.