Evaluation of UNHCR's Repatriation Operation to Mozambique

EVAL/02/96

SCOPE OF THE REVIEW

After consultations between the Regional Bureau for Africa and the (then) Central Evaluation Section, the following terms of reference were agreed for an evaluation review of the Repatriation or Movement Phase of the Mozambique Repatriation Operation, to be supplemented at a later stage by an evaluation of the Reintegration Phase:

• The overall objectives of this part of the evaluation will be (1) to examine the extent to which lessons learned from previous related evaluations were incorporated into the planning process for the Mozambique repatriation, (2) to identify additional lessons learned from this operation, and (3) to examine the relevance of those lessons to the planning and implementation of repatriation programmes elsewhere in the world.

• The review will examine, inter alia, the range of repatriation methods used in this operation, including spontaneous, semi and fully organised return, as well as different modes of travel, including train, truck and boat. Cost-effectiveness of the various options will be reviewed, taking into account the political parameters.

• Original planning assumptions will be assessed against actual accomplishments, in particular with regard to the effectiveness of efforts made to minimise investment in logistics in favour of productive expenditure on initial reintegration activities. Linkages made between movement and initial reintegration, in the both planning and implementation phases, will be also be examined.

• The operation will also be examined from a structural perspective, focusing particularly on linkages between Headquarters and the field, links between countries of first asylum and the country of origin, as well as regional coordination. The use of human resources in this operation will also be assessed.

The terms of reference were referred to and accepted by the Branch Offices in the region, one of which, BO Harare, suggested that the following areas of concern also be addressed:

• Review of the VRF and its use.

• Evaluation of tripartite agreements and negotiations leading to these agreements.

• Registration for voluntary repatriation

• Computerization of registration results

• Information campaign

• Mine awareness campaign

• Durable solutions for especially vulnerable individuals

• Post-creation and deployment of staff

It was further agreed by BO Maputo1 that the evaluation of the repatriation phase should concentrate on countries of asylum 'without leaving Mozambique out entirely'.

The evaluation was conducted through discussions with the Coordinator and other staff members at UNHCR Headquarters and by two field missions, to Tanzania and Malawi between 11 and 23 March 1995 and to South Africa, Mozambique and Zimbabwe between 18 April and 8 May 1995. Both missions involved extensive field visits as well as interviews with government officials and NGO representatives as well as UNHCR staff. Cooperation and hospitality was excellent throughout and the advice and guidance received was, needless to say, invaluable.

All the terms of reference set for the review have been addressed - some, inevitably, more exhaustively than others - in the course of this report. A number of proposals and suggestions have been made which, it is hoped, will assist in the planning and implementation of future repatriation operations.

EXECUTIVE SUMMARY

(1) The main thrust of this report is contained in separate chapters on each of the six countries of asylum - Malawi, South Africa, Swaziland, Tanzania, Zambia and Zimbabwe - and on the country of origin of the refugees, Mozambique. The rationale for considering each country separately is explained in paragraphs 10 to 23 below.

(2) In the course of the evaluation study, however, a number of more general issues were identified for separate examination, namely the Plan of Operation, Coordination and Management and Fund Raising. The conclusions reached on each of these subjects have been briefly summarised below along with those on each of the seven countries concerned with the repatriation.

(3) In general, the outcome of the evaluation amply confirms the accepted view that this was a highly successful operation which reflects great credit on all concerned. Any criticisms made must be viewed in this light and are, in any case mostly country specific as well as relatively minor by comparison with the tremendous achievement of the operation as a whole.

(4) In terms of 'lessons learnt', the main areas can be briefly stated thus:

(a) The appointment of an overall coordinator is fully justified in the context of a complex repatriation operation of this kind but is likely to be most effective if the coordinator is given executive directing powers.

(b) There are arguments in favour of replacing the traditional division of responsibility within UNHCR between the respective Branch Offices in countries of asylum and the country of origin by a system of unified control over the actual repatriation (but not the reintegration) process.

(c) The pace of organised repatriation should reflect the absorptive capacity of the country of origin and be determined accordingly rather than on the movement capability of the individual country of asylum.

(d) Careful and extensive forward planning is a vital ingredient in the success of operations of this kind and has a major role to play in creating donor confidence as well as in ensuring sound implementation.

(e) A systematic, informative and credible fund-raising strategy which is responsive to donor requirements and keeps donors fully informed of operational developments and budgetary changes is crucial to the successful funding of a repatriation programme.

(f) Staff must be deployed to the field in sufficient numbers and in good time before operations commence. Most operational failures and omissions are the direct result of an inadequate staffing response; normal recruitment procedures are too lengthy and cumbersome and emergency procedures may be required.

(g) Repatriation programmes are managed most effectively when operational decisions are taken in the field, as near as possible to the actual site of operations.

N.B. This principle applies to virtually all field operations and programmes, not merely those concerned with repatriation.

(h) UNHCR should aim at more consistent, and therefore more credible, repatriation statistics even if total accuracy cannot be achieved.

(i) An assured and adequate food distribution on return to the country of origin is the single most important item of relief assistance in a repatriation programme of this kind. Sporadic failures in the food distribution system, while understandable in the light of the immense logistics difficulties encountered, were probably the programme's greatest operational deficiency from the refugees' point of view.

(j) Some phases of the operation were probably unduly rushed in order to meet target dates or political deadlines, such as national elections. While it is commonly said that the humanitarian should follow the political agenda, this should not be at the cost of preparedness in the country of origin for the refugees' return.

(k) In the interests of full accountability and equal treatment of implementing partners, there should be no exemptions from compliance with the mandatory standard clauses of UNHCR operational agreements and the waiver granted to IOM in 1992 should be withdrawn.

(l) The proper disposal of assets and, to the extent possible, the restoration of the environment in the country of asylum, should be integral features of any repatriation programme.

(m) An advisory Voluntary Repatriation Unit should be established at Headquarters to prepare and assist with the implementation of standard procedures and guidelines for repatriation operations in general.

These lessons and others are elaborated in the main body of this report. The following paragraphs summarise the report's main conclusions on both general and country-related issues.

The Plan of Operation

(5) Every stage of the operation benefited greatly from the results of careful and extensive planning at both Headquarters and field level. The success of the operation clearly demonstrates the effectiveness of the planning process. The only major component of the Plan of Operation which had to be abandoned was the concept of a travel grant for self-certified return; in all other respects, it was remarkably accurate in its analysis and forecast of events. Nevertheless, the detailed implementation of the Plan showed wide divergences of approach between different countries of asylum which might have been avoided had more power to secure greater uniformity been vested in the Coordinator.

(6) The May 1993 Appeal, which derived directly from the Plan of Operation, was a major factor in engendering donor confidence and in securing funds for repatriation over the three-year period of the Appeal.

Coordination and Management

(7) Within the overall framework of the Regional Bureau, it would have been better for the Coordinator to have had executive authority to direct the operation. It would also have been logical to override the traditional UNHCR division between the responsibilities of Branch Offices and make a single authority - the Coordinator - responsible for the management of the repatriation operation from refugee camps to final destination.

(8) A considerable body of technical competence in the mechanics and logistics of repatriation has now been developed and should not be lost. The creation of a small advisory Voluntary Repatriation Unit at Headquarters would help to preserve the knowledge and experience accumulated over successive repatriation operations in the past and would provide a focus for the planning and administration of future repatriation programmes.

Fund Raising

(9) The excellent donor response to funding needs for the Mozambique repatriation can be attributed in large measure to the systematic, informative and credible fund-raising strategy which was adopted for the operation. Donors were regularly informed of operational needs and budgetary changes, a process which could be developed even further in the future. There was excellent cooperation between the Fund-Raising Service, the Coordinator, the Regional Bureau, the Public Information Section and the Field Offices concerned and this contributed greatly to the quality and quantity of information with which donors were supplied.

Country Summaries

(10) The repatriation to Mozambique was, in a number of respects, unique. It involved the return of some 1.7 million refugees from no less than six adjacent countries of asylum over a time span stretching from late 1992 to the middle of 1995. The number of refugees for whom UNHCR was responsible in individual countries of asylum varied greatly, from over a million in Malawi to figures in the region of twenty thousand in Swaziland, Tanzania and Zambia. The number in South Africa was never precisely ascertained.

(11) The methods adopted by UNHCR for the physical repatriation of refugee populations ranged from road and rail to lake crossing and voyage by sea. Even air transport was at one stage contemplated for vulnerable groups returning from Zimbabwe but was never actually used. The great majority of the refugees, principally from Malawi but also from every other country of asylum to a greater or lesser degree, returned spontaneously by the simplest method of all - on foot.

(12) Before repatriation, refugee communities in countries of asylum had been accommodated in camps in agricultural settlements or had settled spontaneously among local population groups. UNHCR assisted refugees in all three of these categories to return to Mozambique, either by the organized repatriation of entire camps or settlement populations or of vulnerable groups, including those who were too far from home to travel on their own. Another method used, especially in Malawi, was so-called 'assisted spontaneous repatriation' which generally consisted in making transport and nothing more available for those who wanted to go home.

(13) Despite a number of common elements such as the machinery of the Tripartite Commission and the use of the Voluntary Repatriation Form or VRF for registration purposes, each Branch Office adopted its own repatriation systems and methods. The VRF, although more than adequate for a wide range of purposes relating to the reabsorption of returnees in their country of origin, was mostly used - except by Branch Office, Harare - simply as a travel document and as evidence of eligibility for assistance on return. The size of the repatriation 'package' varied from country to country as did logistic arrangements and the use of computerised techniques. Given the great differences in the repatriation environment between countries of asylum, any attempt to impose greater uniformity would probably have been unwise. The length of the operation also varied greatly from country to country - the entire period of (roughly) three years in the case of Malawi at one extreme or only three months in the case of Tanzania at the other.

(14) All these and other disparities have meant that, for evaluation purposes, it has proved impossible to treat the operation as a single unified whole. Repatriation from each country of asylum has been treated separately in its own country chapter and these together with a chapter on the country of origin, Mozambique, have been grouped together as a distinct part of this report. Each country chapter contains, however, a good deal of detail of special interest to the relevant Desk at Headquarters and the concerned Branch Office but which the more general reader may not wish to follow. The ensuing summaries are therefore aimed at providing a very brief statement of the principal lessons learnt from each country's share of the operation; a fuller statement can be found in the 'conclusions' section of each country chapter.

Malawi: 1992-95 1 million+ repatriated

(15) The Branch Office did excellent work in facilitating, promoting, organising and assisting voluntary repatriation. Nevertheless, it was the refugees themselves who played the predominant role and contributed most to the operation's success.

The policy of 'food consolidation' was an inevitable necessity if the 'revolving door' syndrome was to be avoided. But the strict criteria for the policy - an assured and regular food distribution on the Mozambique side - were not always met.

(16) The country of origin, not the country of asylum, should dictate the pace of any repatriation which UNHCR has the power to regulate. This important principle was not always observed, particularly in relation to 'assisted spontaneous repatriation'. In general, however, cooperation and coordination between UNHCR Malawi and UNHCR Mozambique were good, especially at field level and any tensions which arose were a natural consequence of different concerns at the repatriating and receiving ends of the operation.

South Africa January 1994 - April 1995 32,010 repatriated

(17) On its own terms, a highly successful operation which helped to create a secure base from which the UNHCR Regional Office can operate in future. Nevertheless, there were unsatisfactory elements - the size of the caseload was never clearly ascertained, the numbers repatriated fell far short of expectations and at the end of the exercise no one could say how many refugees were left in South Africa. Only the roughest test of eligibility could be applied.

Swaziland October 1993 - June 1994 16,000 repatriated

(18) A well-conducted and successful operation mainly by railway and with relatively few logistic problems. The 'repatriation package' was particularly generous as was the baggage allowance but these may have assisted greatly in reconstruction in Mozambique. The statistics relating to refugee numbers in Swaziland are -even by UNHCR standards - unusually confused.

Tanzania June 1994 - October 1994 11,400 repatriated

(19) As an exercise in orderly and organized repatriation in difficult logistic circumstances, the operation was an unqualified success. There were, however, delays in providing essential staff and equipment and UNHCR was fortunate in having an experienced and efficient NGO presence to rely on. It can also be argued that, given the extremely fragile infrastructure on the Mozambican side, the timetable was too rushed.

Zambia October 1993 - December 1994 19,927 repatriated

(20) A smooth, successful and well-conducted operation, greatly aided by a strong NGO presence.

Zimbabwe June 1993 - May 1995 117,702 repatriated

(21) Another extremely successful operation, meticulously planned and executed with greater use than elsewhere of computerised registration and related techniques and special consideration for vulnerable groups. However, the repatriation of the spontaneously settled Mozambican refugees in Zimbabwe, which followed that of those resident in camps, was advertised and carried out over an extremely short time span, mainly in order to meet the wishes of UNHCR in the country of origin. A longer and much less hurried operation might have given more refugees the opportunity to go home.

Mozambique 1.7 million received and reintegrated

(22) UNHCR Mozambique had to undergo a challenging transformation from a small Branch Office with a single modest Sub-Office to a major operational centre with three Sub-Offices and field offices in all the main areas to which repatriating refugees returned. The transition - through no fault of the Branch Office - took far too long and not until mid-1994, when half or more of the refugees had already come back, were staffing requirements reasonably met. Resources were often strained to the limit and emergency staffing procedures would have been justified.

(23) Despite many logistic problems and a fragile governmental infrastructure, every effort was made to provide adequate reception facilities and to restore basic services after many years of civil war. Throughout the repatriation operation, the Branch Office sought to assert the principle that the country of origin must determine the rate of repatriation (other than unassisted spontaneous return, over which UNHCR can obviously have no control). This principle was generally observed, but some difficulties arose, particularly in the case of "assisted spontaneous" repatriation from Malawi.

THE PLAN OF OPERATION

(24) The origins of a planned approach to the repatriation of Mozambican refugees are summarised in the Introduction to Part I of the May 1993 Plan of Operation. The first major initiative was a joint UNHCR/UNDP/Norwegian Refugee Council (NRC) mission to Mozambique in March 1992 to begin preparations for the basic framework of the operation. The mission was funded by the Norwegian Government and was said to be the first of its kind in UNHCR's history, embracing as it did participation by representatives of the donor community, the NGO movement, bilateral aid agencies and the UN system. Be that as it may, for the actual composition was a good deal narrower that this might imply, the mission did recommend an integrated and coordinated approach to the operation, a conclusion which was endorsed at the so-called Nordic Conference in Oslo on 22 June 1992. The mission also defined the main planning parameters to be addressed from target groups to implementing mechanisms and funding strategies.

(25) Few things are easier, of course, than to bandy over-used words such as 'integrated', 'coordinated' and 'strategy'. Nevertheless, the Oslo resolutions did, as the Plan of Operation pointed out (para.8) provide a real start in the planning process and led directly to the preparation of the policy framework document presented to the Donors' Conference in Rome in December 1992. This in turn provided the basis for the overall plan which was drafted at a regional workshop on repatriation held in Malawi in late January 1993 and which finally appeared in published form, along with an Executive Summary, in May 1993. Simultaneously with the publication of the Plan, the Fund Raising Service issued its first Appeal for the Repatriation and Reintegration of Mozambican Refugees, dated 3 May 1993, drawing heavily on the Plan of Operation and calling for contributions of US$203.4 million over the three year period 1993-95.

(26) 18-21 January 1993, shortly before the January 1993 regional repatriation workshop, a regional meeting of protection officers was also held in Maputo to consider a draft Protection Plan of Action and to adopt a standard approach to the protection issues which the operation would have to face. The principal elements of the protection plan were incorporated into the legal framework and provisions of the Plan of Operation, covering such fundamental questions as Tripartite Agreements, promotion of and registration for voluntary repatriation, Voluntary Repatriation Forms (VRFs), the right to return, property rights and physical security. Reading the draft protection plan long after the event, it seems an admirable attempt to identify problems and provide a basis for generally applicable procedures and solutions.

(27) However, the protection plan was not without its critics. One Branch Office in particular - Zimbabwe - expressed disappointment at what it considered a failure to address protection-related issues and was especially critical of the draft Tripartite Agreement which it said 'leaves more issues open for dispute than it provides for mechanisms and solutions for very likely problems'.2 The Branch Office had suggested a much more detailed document covering a number of major operational issues and submitted its own version to Headquarters on 16 December 19923. Nevertheless, it was eventually obliged to accept a Headquarters' version and this was signed between the three parties on 22 March 1993. It was identical in all respects save one - provision for refugee representatives to visit Mozambique, which was omitted - to the specimen Tripartite Agreement which subsequently appeared as Annex II to the May 1993 Plan of Operation. A proviso to the conclusions of the January 1993 meeting of protection officers that Tripartite Agreements could incorporate country specific requirements was therefore not invoked.4

(28) The genesis of the overall Plan of Operation for the repatriation of Mozambican refugees can also be traced to a meeting in Geneva, to which the Branch Office Representatives from Malawi, Mozambique and Zimbabwe were invited, between 18 and 21 May 1992. The main purpose of the meeting was to consider the implications for UNHCR programmes of the severe drought in southern Africa but among the other items on the agenda was "Repatriation: Preparatory Arrangements before mass repatriation".

(29) The meeting commented that earlier mass repatriations - Ugandans from south Sudan in 1986, Ethiopians from Somalia in 1991, Sudanese from Ethiopia in 1991 - had developed unnecessarily into emergency situations because UNHCR had ignored the warning signs and waited until the last moment to act. It urged immediate action to prepare for repatriation in both Mozambique and asylum countries including the following range of activities:

1. Data collection on the refugee caseload, areas of origin, distance from the border.

2. Transport and logistics requirements.

3. State of roads, especially in Mozambique.

4. The need for de-mining.

5. Identification of entry points, transit and reception centres.

6. The role of other UN agencies and NGOs.

7. Coordination between Branch Offices - the design of a comprehensive Plan of Operation.

8. The need for Tripartite Agreements, similar to the Malawi one and the strengthening of the Mozambique Branch Office.

(30) Resource needs were identified: Durable Solution teams in each Branch Office, a major purchase of radio/telecommunications equipment, extensive regional travel by repatriation staff (to be authorised by Representatives, not Headquarters) and the appointment of a Regional Coordinator within the Desk 'at the appropriate time'.5

Finally, mention should be made of the full country chapter plans of operation which each Branch Office prepared and which are referred to in paragraph 12 of the Plan of Operation. These are impressive documents - that for Malawi, prepared jointly with Branch office Mozambique goes back to 1990-91 and has 27 substantive pages with 28 appendices. Branch Office Mozambique subsequently produced its own 33 page Contingency Plan covering repatriation from all countries of asylum except South Africa.6 The Zimbabwe plan of March 1993 covers every aspect of repatriation during the registration, pre-departure and movement phases in a comprehensive document of 54 pages.

(31) These country plans were obviously too long to be included in the overall Plan of Operation but constituted the basis for Part II of the Plan - Country Activities - in which the programme needs of each country of asylum and the country of origin were summarised, except in the case of South Africa where only a country profile could be provided at this stage. They must have contributed greatly to the wider picture and to the practical and realistic nature of the Plan.

CONCLUSIONS

(32) The Plan of Operation and the Executive Summary and Appeal derived from it were clearly major assets in persuading donors that repatriation to Mozambique was an activity worthy of their support. The fact that funding the operation was never a problem is itself a tribute to their worth. The Plan was also remarkably accurate in its analysis and forecasts of events and in the time frames which it suggested (paragraph 50) for operations in the different countries of asylum.

(33) The only significant ingredient of the Plan which - mercifully, as it turned out - did not materialize was the proposal for self-certified return with a travel grant for refugees falling within the organised repatriation operation but able to return home without transport assistance from UNHCR. This category was expected to come principally from Malawi (para.32 of the Plan) and to depart with the expiry of care and maintenance assistance (para.66).

(34) The travel grant would include components to cover transport costs for the refugees and their belongings to their destination in Mozambique and 'costs for overnight stays and food during the journey'. How these would be calculated was not elaborated in the Plan but a glance at the relevant paragraphs, 93 to 99, is sufficient to demonstrate how complicated the arrangements would have been and how much nightmarish accounting might have been required. For budgetary purposes, Malawi Kwacha 100 per head of family and 50 Kwacha per family member was estimated (para.164) for Malawi and an average of US$27 per person for Zimbabwe (para.202). Zambia did not anticipate using certified self-repatriation (para.182) to any extent and Tanzania did not specify what cash grant would be paid but said that the practice of refugees making their own transportation arrangements would be encouraged.

(35) Presumably the idea of a cash grant for self-repatriation derived from the generally positive experience of a similar system for Afghan refugees in Pakistan which was introduced in 1990 and reached its peak in 1992. A later evaluation of the Pakistan programme7 concluded that 'encashment has served as an effective catalyst for large-scale spontaneous repatriation' and had 'demonstrated the potential for using cash to support refugee choice'. This evaluator can only record his opinion that, in the Mozambican context, it would have been a disaster. Quite apart from the complicated administration involved, the introduction of cash grants would very probably have killed spontaneous repatriation in the sense of unassisted return, since what refugee would want to take himself and his family home without assistance once he knew that UNHCR would pay him to do so. The result would have been an enormous increase to no useful purpose in programme costs.

(36) Fortunately, this was quickly realized8 and the authorities in countries of asylum e.g. Zimbabwe, also objected to the scheme. By the time of the January 1994 Progress Report, UNHCR had concluded that 'the system of paying cash grants to refugees in order to encourage them to repatriate may not be suitable for this operation'. A major planning mishap had been averted.

(37) In retrospect, it could also be argued that the Plan, while acknowledging that spontaneous repatriation would continue, underestimated its extent. But this is a pointless criticism, wise only after the event. Also in retrospect, the argument over the terms of Tripartite Agreements seems a storm in a teacup, since the atmosphere in which repatriation took place was cooperative throughout all countries of asylum and in the country of origin and no significant disputes took place. This is not to say, however, that there is not a case for greater particularity in individual country circumstances and some of the ideas in the Zimbabwe draft were definitely useful - such as, for example, recognition in the country of origin of the juridical status and qualifications of refugees. But the divergencies from the accepted model were not great and there are also arguments in favour of a generalised and even vague approach. The golden rule, as in so much else, is that there is no golden rule.

(38) In conclusion, there is one point of overriding importance to be made. Few, if any, earlier repatriation operations can have benefited from so much meticulous and extensive planning at both field and Headquarters' levels.9 The fact that, as the three Representatives who came to Geneva in May 1992 had urged, an emergency approach was avoided and that - with only minor hiccups - the operation was so successfully conducted is a direct tribute to the effectiveness of its planning. This was no mere public relations exercise; the beneficial results of careful planning showed themselves at every stage and reflect great credit on all concerned. It is heartening to see that the same attitude is being taken towards the operation for repatriation to Angola.

(39) One final note of caution. One Branch Office has pointed out that, whereas the regional standardisation of such aspects of the operation as registration, computerisation, pre-departure assistance, the needs of vulnerable groups, mine awareness and others was emphasized during the initial planning stage and the Plan of Operation contained many guidelines towards a regional approach, there was in practice little follow-up and Branch Offices developed their own individual policies and methods. This led to wide divergencies and little uniformity between the six countries of asylum. In the opinion of the evaluation mission, this might have been avoided had authority been vested in the Coordinator to ensure that, to the extent possible, regional uniformity was achieved.

(40) But standardisation for its own sake or for the appearance of tidiness serves no purpose and should only be applied when real benefits are likely to result. It is hard to be confident that greater standardisation would have led to a more successful operation.

COORDINATION AND MANAGEMENT

(41) A Coordinator for Southern Africa was appointed within the Africa Bureau in September 1992, specifically to coordinate the repatriation and reintegration of refugees returning from neighbouring countries to Mozambique. Apart from this, there were no precise terms of reference, no line management responsibility, no executive authority and consequently no special unit in support. The Coordinator had no staff, apart from a secretary who arrived four months after his appointment. His principal operational links were with Desk V within the Bureau, which covered both the country of origin and all the countries of asylum save Tanzania and, in the case of Tanzania, with Desk IV.

(42) This is a familiar pattern and one which has in the past proved a recipe for disaster. That it was far from a disaster in this instance, but rather a considerable success can be attributed to three factors. In the first place, the Coordinator's lack of authority was at least mitigated by the fact that his area of intervention - the Mozambique repatriation and reintegration - was clearly defined. He was fortunately not merely given a vague but all-embracing brief to 'coordinate' UNHCR activities in a given geographical zone, a task which of its nature invites defeat, but was at least given a precise sphere of activity within which to function. In the second place, his personal attributes of drive, enthusiasm, diplomacy and persuasive power proved more than equal to surmount the challenge. His was a constant presence, travelling, exhorting, accelerating, reconciling, exercising in the best possible way, as Walter Bagehot said of the British monarchy in happier times, "the right to be consulted, the right to encourage, the right to warn". In the third place the Bureau, the Desks and the Branch Offices in the region all did their utmost to strive for the operation's success. Moreover, key staff within Desk V had invaluable recent experience in the region and of repatriation operations elsewhere.

(43) The fact remains that, in management terms, to ask someone to coordinate an operation without operational authority is a contradiction and is likely, in most circumstances, to lead to confrontation when he (or she) exceeds his mandate, and ineffectiveness when he does not. Far better, if a coordinating role is considered necessary - as it certainly is when repatriation from six countries of asylum is involved - to appoint someone with genuine executive powers over the operation within the overall framework of the Bureau. This in turn implies the allocation, where necessary, of supporting staff and of authority vis-à-vis Representatives in the field10.

(44) This issue of authority is intrinsically of considerably more importance than the question, which seems to have been extensively debated, whether the post of Coordinator should have been established at Headquarters or in the field. The answer to the one question probably turns, in any event, on the solution to the other. For a Coordinator without authority, the best central forum may well be Headquarters with - as has been the case in the Mozambique operation - frequent missions to the field. Assignment to a field duty station (and which, with seven to choose from?) might only serve to exacerbate any possible source of friction, always a danger when respective roles and responsibilities are ill-defined.

(45) On the other hand, for a Coordinator or equivalent with executive powers and on the obvious and well-tested management principle that the closer an enterprise is administered to the site of operations the better managed it will be, a field assignment is more appropriate. Had this principle been adopted in the case of Mozambique, it would have been preferable to establish the Coordinator's presence at a centre with a good regional communications network but away from either of the main capitals involved, i.e. not in Maputo or Lilongwe, so that neither of the principal actors would feel that the Coordinator was 'breathing down his neck'. The Regional Office in Johannesburg, had it been created in time, would have been an obvious possibility or even, for the closest operational involvement, Blantyre in Malawi - the nearest practical central point to where the great majority of the refugees had sought asylum and to where they would return.

(46) The existence of a Coordinator with executive authority implies the power to supervise and in the last analysis, to overrule individual Representatives on the overall direction of the programme, at least so far as the process of repatriation is concerned. After a great deal of thought, this evaluator has come to the conclusion that in mounting a repatriation operation, the traditional division in UNHCR between Branch Offices can be more of a hindrance than an asset. The reasoning is clear. Repatriation is essentially a single logistic process, involving the voluntary movement of large numbers of people from a temporary residence to their former permanent place of abode. There is no good operational reason why overall responsibility for that process should be divided between two sets of people, whether UNHCR or implementing agency personnel, once the national boundary is crossed between the country of asylum and the country of origin. There is implicit recognition of this fact in the incorporation in Tripartite Agreements of provision for simplified border procedures, cross-border travel authorisations and the like.

(47) But this recognition does not go far enough. The fact remains that, in UNHCR, responsibility shifts at a certain reception point along the road from one authority to another in a manner which is quite irrelevant to the needs of the individual repatriating refugee, who only wants to be taken in safety and dignity to his or her final destination. The best way surely to ensure that the refugees' needs are properly met at every stage of the operation is to place it under a single unified control. This requires in turn that field staff on both sides of the national boundary are responsible to a single managerial authority - the Coordinator - for all aspects of the movement from refugee camp to final destination.

(48) This concept needs, of course, to be thought through a good deal more thoroughly than is possible in this report. It does not remove the responsibility for close liaison and consultation between the Coordinator and the Representative in the country of asylum, who remains responsible for all aspects of the refugees' welfare other than those relating to the process of repatriation, from registration to departure. It does not impinge on authority for reintegration, an ongoing process which is clearly the responsibility of the Representative and the national authorities in the country of origin. It is not so very different, in terms of practical realities on the ground, from what has in fact been happening through close cross-border cooperation between UNHCR field personnel on both sides of national frontiers.

(49) There has been, nevertheless, a good deal of tension between the UNHCR Branch Offices in the country of origin, Mozambique, and the main country of asylum, Malawi. This is common knowledge and has been commented on elsewhere. Whatever other reasons there may be for the difficult relationship between the two offices, one must surely be that, on the issue of repatriation, the interests of the country of asylum and the country of origin can never be the same. However much both may genuinely be seeking to attain a common goal, there is an inevitable element of burden-shifting involved. For the country of asylum, a refugee repatriated is a problem solved; for the country of origin both in terms of short-term relief and longer-term rehabilitation and reintegration, it is an added problem to be tackled. The refugee moves, so to speak, from the one Representative's 'out' tray to the 'in' tray of the other. The more reason, therefore, to put the process of repatriation under an independent, impartial third command able to operate equally and impartially on both sides.

(50) Two other management points. In all of the five countries visited during the evaluation mission, the standard of performance, energy and dedication of international, national officers and local staff has been most impressive as has that within the Bureau at Headquarters. From a comparative, if subjective, perspective, it seemed significantly higher than would have been the case ten or twenty years ago. The high esteem in which UNHCR is held generally in the region supports this view. A considerable body of technical competence in the mechanics and logistics of repatriation has been built up which should not be lost.

(51) The second point follows from the first. Repatriation has become a technically sophisticated and complex process, involving difficult logistic and transportation problems, computerised registration, detailed manifests, advanced communication systems, intricate liaison mechanisms with police, immigration customs and other services and much besides. As the most desirable of all durable solutions and with new challenges, such as Angola, directly ahead, it would surely be worth creating at Headquarters a small, advisory Voluntary Repatriation Unit which would assist with the operational planning of repatriation programmes and, if necessary, with the setting-up of appropriate administrative mechanisms in the field. Such a Unit could be established either in the Division of International Protection or the Division of Programmes and Operational Support - (perhaps the latter, since the key function is operational preparation and support rather than legal advice although this would not be excluded). The necessary expertise to staff it is readily available from those who have so successfully planned and implemented the Mozambique repatriation.

(52) Some of the many issues with which the Voluntary Repatriation Unit would be concerned are:

1. Tripartite Agreements and associated procedures covering amnesties, guarantees of safety, immigration and customs formalities, cross-border travel authorisations, training of national officials, and family reunion, transit centres.

2. Promotion/information campaigns (radio, leaflets, meetings, information on the country of origin)

3. Transport and logistic methods

4. Telecommunications systems

5. Voluntary Repatriation Forms, content, computerisation, usefulness in country of origin (their use in the Mozambique context was mostly limited to providing documentation for travel purposes and evidence of eligibility for assistance in Mozambique).

6. Strategies for dealing with vulnerable groups.

7. Baggage entitlement, the provision of food, family kits, currency exchange procedures.

8. Health arrangements - vaccination etc.

9. Phase-out, handover and environmental restoration in countries of asylum.

10. Staffing, training, team-building in the field.

11. Mine awareness campaigns

12. Reintegration measures in countries of origin (schools, roads, clinics, water supplies, agriculture, community services, NGO participation, capacity building for government services and institutions).

The aim would be, as far as possible, to prepare standard repatriation 'packages' and guidelines which, with modification to suit local conditions, could be used speedily and effectively whenever a new repatriation operation was envisaged.

(53) Finally, a plea nevertheless for flexibility in meeting the organisational needs of future repatriation operations, changing or adapting the structure to meet the situation rather than trying to make the situation conform to an existing structure. And remembering always Alexander Pope's dictum of two hundred and fifty years ago - "For forms of government let fools contest, whate'er is best administered is best".

FUND RAISING

(54) UNHCR's fund raising strategy for the Mozambique repatriation was based, to a far greater extent than had been the case in many earlier funding situations, on the preparation and dissemination to donors of a comprehensive and detailed operational plan. This was based on the many drafts which Branch Offices in the region had prepared, on the framework document prepared for the Donors' Conference on Mozambique in Rome in December 1992 and on a final draft prepared at the regional workshop on repatriation held in Malawi in January 1993. The regional workshop also took into account the conclusions of a regional meeting of protection officers which had taken place in Maputo earlier the same month. The entire planning exercise lasted at least a year.

(55) The result was a 56-page (excluding annexes) Plan of Operation which, together with a five page executive summary, formed the supporting documentation for a formal appeal to donors issued on 3 May 199311. Although the initial appeal was for a 1993 requirement of US$55 million, it was made clear that the operation would have a three year span up to 1995 and a breakdown of the overall financial requirement of US$203 million was provided by year as well as by country and sector of activity, the only provisional budget being one of US$12million for South Africa, pending a comprehensive assessment. It was also made clear that needs would be revised to reflect operational realities and, indeed, in September 1993 an update was issued reducing the estimated need for 1993 to US$45 million of which US$31 million had already been received. Despite the reduced estimate, however, US$51.5 million was contributed during 1993 (including a carry-over of US$3.4 million from 1992).

(56) The message to donors was, therefore, that, in return for a long-term commitment to a massive operation requiring separate annual budgets for the country of origin and six countries of asylum, they could be assured that these budgets had been realistically assessed and would be regularly and realistically reviewed. Problems and delays would be admitted12 and progress fully reported both by UNHCR Headquarters and in the field. Indeed, UNHCR Mozambique published its own Special Newssheet at intervals from September 1993 onwards as did UNHCR Malawi with regular updates under the headline 'ZIKOMO MALAWI'. Donor representatives in key countries in the region were given a systematic and ample flow of information to enable them to report comprehensively to their capitals, with probably as much and possibly even more influence than their counterparts at diplomatic missions in Geneva. Be that as it may, however, the Fund Raising Service in Geneva took the trouble to organise in 1994 a mission to Malawi and Mozambique of a number of donor representatives including the head of Canadian CIDA and the senior US State Department official for Southern Africa, something which it appears had never been done before13.

(57) A full Progress Report on 1993 and programme projections for 1994 was published in January 1994 and distributed on 8 February with the 1994 Appeal, now totalling US$102.8 million - a substantial increase over the figure of US$82 million originally forecast for 1994 at the time of the May 1993 Appeal. Surprisingly perhaps, the 1994 appeal did not contain any explanation for the increase of over US$20 million, relying presumably on the earlier warning that needs would have to be revised. The 1993 Plan of Operation did not contain any detailed breakdown of projected annual expenditures by sector or country, whereas a brief account of the main areas in which funding requirements had increased might have helped donors to a better understanding of how operational realities had evolved. This is not a criticism but a suggestion as to how the tremendous strides already taken in the direction of detail and transparency might have been taken a step further.

(58) The High Commissioner returned from a two-week visit to Mozambique, South Africa and Zimbabwe in mid-March 1994 and wrote immediately to donors urging their support for the 8 February Appeal, backed by the unequivocal statement that "I am greatly encouraged by what I have seen". There could hardly have been a more powerful endorsement, an opportunity eagerly taken and exploited to the full.

(59) Two further Progress Reports on the operation were issued in June 1994 and January 1995 respectively, although neither referred directly to funding needs. However, the January 1995 report was published concurrently with the 1995 Appeal which now extended the programme to a final conclusion of both repatriation and reintegration operations in June 1996. It called for the provision of US$56.9 million over an eighteen-month period, of which by far the greater share - US$45.6 million - was for the first time, to be spent on reintegration activities in Mozambique itself.

(60) But the 1995 Appeal also indicated that 1994 contributions had totalled only US$73.4 million, including a carryover of US$13 million from 1993 and cancellations of US$6.6 million. Actual donor contributions amounted to only US$53.9million, an apparent shortfall of almost US$50 million (or almost half) of the overall requirement of US$102.8 million or nearly US$30 million if the carry-over and cancellations are included in the contributions figure. Nothing in the Appeal or its associated documentation suggested any adverse consequences from the shortfall and no scaling-down of requirements had been announced.

(61) In February 1995, the Public Information Section at Headquarters issued an excellent Special Report "Mozambique - Back on Track" showing donor contributions for 1994 at US$53,787,391 against a 1994 Budget of US$51,071,192 implying a reduction in 1994 needs of almost exactly half since the 1994 Appeal and, in the light of such a substantial cut, no funding shortfall but instead a modest surplus of actual donor contributions during the year. The Special Report also gave a 1995 budget figure of US$42.7 million, whereas there is no breakdown between 1995 and 1996 in the 1995 Appeal.

CONCLUSION

(62) All this is a little perplexing and suggests that more still needs to be done to make the donors fully aware - at least on paper - of the reasons for major budgetary changes and their implications for the programme. Documentary material was, of course, supplemented by regular oral briefings of donor representatives both at Headquarters and in the field so that adjustments and budgetary details may well have been explained in the course of these briefings.

(63) The fact remains that, in terms of adjusted requirements, the funding response to the Mozambique repatriation and reintegration operation has been exemplary14 at a time of exceptionally heavy Special Programme needs elsewhere (notably in the former Yugoslavia). This can be accounted for not only by the inherent popularity of durable solution programmes, especially voluntary repatriation, but also by the sustained, coordinated, honest and informed fund raising campaign which has underpinned the entire operation. The quality of this campaign reflects a very high level of cooperation between the Fund Raising Service, the Regional Bureau, the Coordinator for Southern Africa, the Public Information Section and the concerned UNHCR Field Offices in the region.

(64) The same approach is being used for the repatriation and reintegration of Angolan refugees for which a 30-month appeal for the period up to December 1997 was issued in June 1995. The appeal document states, on page 12, that 'donors will be regularly informed of operational developments and the inevitable refinement of project planning and budget revisions as work proceeds'.

(65) The lessons learnt from this evaluation are (I) a systematic, informative and credible strategy, with clearly stated performance targets linked to budgetary needs, was developed for the Mozambique operation and is already becoming a model for programmes of this kind, and (ii) the value of this strategy could be enhanced even further if the inevitable refinement of project planning and budget revisions were more fully explained when major changes in budgetary needs take place. The prime source of the explanations should be the Fund Raising Service.

(66) One final happy thought. However one calculates the figures - and the reintegration phase still has nearly a year to run - the operation must have cost much less than the US$203 million originally projected. US$150million would be a reasonable guess.

MOZAMBIQUE

(67) UNHCR first established a presence in Mozambique immediately after independence in June 1975, mainly to handle the first repatriation of more than 500,000 Mozambican exiles and, until their repatriation in 1980, had also to care for a sizeable caseload of Zimbabwean refugees. Thereafter, the Branch Office declined in importance until by 1986 total annual expenditures were significantly less than one million dollars and the staff of the office consisted of only 3 professionals (including one JPO) and 5 GS personnel. A Sub-Office in Tete was opened in 1987 to deal with returnees from Malawi but its capacity for movement was severely restricted because of the civil war and some early plans to accommodate repatriates in so-called 'safer area' settlements never developed significantly. A Field Office was opened in Chimoio, Manica Province in 1988. In spite of these developments, UNHCR Mozambique remained essentially a small office with limited resources until the repatriation operation began in earnest in 1992/93.

(68) Mozambique is a large country covering a total land area of 784,095 square kilometres, smaller than Tanzania but greater than Zambia, Zimbabwe and, of course, Malawi. Even by African standards - the colonial legacy at its worst - it has a ridiculous shape, grotesquely elongated with the capital - Maputo - in the extreme south and 40 percent of the population living in the two provinces of Zambezia and Nampula in the north. Of an estimated population of 16 million, some four million were internally displaced by the drought and by the conflict between the FRELIMO government and RENAMO. The great majority of those displaced were resettled, together with some 200,000 demobilised soldiers, between late 1992 and the end of 1994. By the end of 1994 also, the vast majority of the almost 1.7 million Mozambican refugees in neighbouring countries had returned home.

(69) The Mozambican administrative structure is fragile at best. Responsibility at government level for the reception and registration of returning refugees is vested in UNHCR's counterpart authority, Nucleo de Apoio aos Refugiados (NAR), which first opened an office in Tete in 1987 and from 1992 onwards spread into all provinces, except Nampula and Inhambane where there are very few returnees. By early 1995, it had 200 staff altogether - with financial support from UNHCR - with field offices in returnee-receiving districts and registrars at entry points. Relations with UNHCR have been consistently good.

(70) According to NAR, it now works with about 100 NGOs on returnee assistance programmes of one kind or another. Much of this must be a recent development; according to UNOHAC (the United Nations Office for Humanitarian Assistance Coordination in Mozambique), there were in October 1993, 46 out of Mozambique's 128 districts which had only 0 to 3 NGOs, 42 with 3 to 5 and 40 with 5 to 14. Those with between 0 and 3 included much of Tete and Gaza Provinces, all of Manica and Niassa Provinces and nearly all of Cabo Delgado - all areas with returning former refugees. By early 1994, however, UNHCR itself had signed agreements with over 40 operational partners, including NGOs, for support in the repatriation and reintegration process. Among the international NGOs involved were World Vision International (WVI) in food distribution to returnees, MSF, SCF and others in support of the health system, LWF and others in infrastructural development.

(71) As elsewhere, however, UNHCR's response to the need for increased staffing levels was painfully slow. This was exacerbated in the case of Mozambique by the fact that, even before the October 1992 Peace Accord, the staff of the Branch Office had been inadequate. As repatriation gained momentum thereafter, the situation became desperate. That is how the DHC put it in his mission report of 18 June 1993 - he went on to say "the full component of international and local staff will not be in place till August". This was a far too optimistic forecast. Despite the increases in staffing which had been approved by the PPR in March 1993, the Branch Office was able to state as late as February 1994 that, of 55 international and 202 local posts which had been approved, only 33 international and 84 local posts had been filled. The vital Tete Sub-Office had only two international staff during the first half of 1993, by the end of which over 260,000 refugees had returned to the Province from Malawi, while in districts such as Angonia - which had been denuded during the war years - almost the entire population now consisted of spontaneously repatriated refugees. By February 1994 also, while eight field offices had been established, critical staffing gaps remained. The Sub-Office at Quelimane, which was split off from Tete15, was not opened until May 1994 and a UNHCR presence in Lichinga was not established until February 1994, in Pemba until March 1994, in Mandimba in March 1994 and Morrumbala in May 199416. The Cessna 402 aircraft allocated to Sub-Office Quelimane and which rapidly became an operational necessity, did not arrive until 11 July 1994.

(72) More examples of inadequate or late staffing arrangements could be cited, including that of the Branch Office, where a new Representative did not arrive until late in May 1993 after a hiatus of about a year. The situation in Mozambique was particularly serious in that

(a) a major UNHCR presence had to be established from so low a base and without a nucleus of experience on which to build;

(b) the Branch Office had to deal with the reception and welfare of repatriates from six different countries of asylum and their reintegration into Mozambican society17, whereas the countries of asylum were concerned with the repatriation process only and with such on-going care and maintenance obligations which remained until repatriation was completed;

(c) the governmental and NGO structures were considerably weaker than in any of the countries of asylum, communications were extremely difficult and many areas inaccessible, particularly during the rains. The transportation of returnees and the distribution of food and other assistance items faced constant operational handicaps including, for example, the poor quality and limited availability of trucks;

(d) The Branch Office had progressively to assume responsibility for local administrative systems and units such as a Contracts Committee, Property Survey Board, APPC, Assets Management System and Medical Insurance Plan while its own organisational capacity was still in the process of being developed18.

(e) The official language of Mozambique is Portuguese, making it necessary that a substantial proportion of international staff with representational or field duties to perform be reasonably fluent in that language. This obviously made recruitment more difficult and resulted in the appointment of some staff members with little or no prior African experience (although they appear to have performed most creditably just the same).

(73) In summary, it is fair to say that the Mozambique Branch Office was not adequately equipped with the necessary staff establishment to carry out its repatriation/reintegration task until at least the middle of 1994. By the end of 1993, however, over 650,000 refugees had already returned to Mozambique and the pace of repatriation actually increased during 1994. An Inspection Mission to the region led by Mr Walter Koisser in July 1994 concluded that (page 2 para.9): "it took too long for the organisation to establish the network of field offices required for timely implementation and effective monitoring of repatriation and reintegration activities."

(74) This is, of course, a complaint echoed elsewhere in the region but it became particularly serious in the difficult and conflict-ridden circumstances of Mozambique. Its general nature does lead one to ask, however, whether the emergency recruitment procedures which had been adopted in the crisis situations of the Gulf and the former Yugoslavia might not legitimately have been extended to this, the largest repatriation operation ever in the African continent and probably the most complex on any continent considering the number of countries of asylum, their differing natures and the immense logistic problems to be overcome. While a line has to be drawn somewhere between emergency and non-emergency situations, operations of this urgency and magnitude surely justify an emergency recruitment response, combined with the use of benchmark job descriptions for standard field posts.

(75) Most of activities of the Branch Office, its Sub- and Field Offices in Mozambique have been concerned with the reintegration aspect of the operation or at least with activities, such as the initial distribution of seeds and tools or of food until the first harvest, which lie on the border between repatriation and reintegration. These activities are not directly the subject of this evaluation but they illustrate what has been the Branch Office's main concern of principle throughout the operation and which is referred to elsewhere in this report - namely that the country of origin should always be the focal point of any repatriation, should ultimately control the pace of any UNHCR involvement in the repatriation process19 and that resources should flow from the countries of asylum to the country of origin as care and maintenance programmes in the former diminish in size while operations in the latter correspondingly increase20.

(76) Given the enormous scale of spontaneous repatriation from Malawi in particular to Mozambique, the common acceptance of these principles might not have had much overall effect on the reintegration effort but would certainly have contributed to better understanding between the Branch Offices involved. The most basic elements of the reintegration package were under considerable strain from the beginning - the September 1993 update on Repatriation and Reintegration estimated that "the distribution of food reaches only a limited number of returnees (some 40 percent) and this may even be lower in some of the border areas". In the most inaccessible areas, such as northern Morrumbala district of Zambezia Province or much of the Niassa and Cabo Delgado provinces, the extent of the coverage achieved is problematic although the number repatriated to some particularly isolated parts of Mozambique were relatively small. Nevertheless, up to early February 1995 UNHCR Malawi repatriated to Mozambique in an organised or 'assisted spontaneous' fashion a total of 128,340 refugees, a figure little more than 10,000 greater than the total repatriated from Zimbabwe and, especially since many were vulnerable cases, a reintegration caseload that could not simply be allowed to fend for itself.

(77) In other countries of asylum, notably South Africa and Zimbabwe, the principle of full consultation with the country of origin over numbers was always accepted in relation to the repatriation of spontaneously settled refugees, a procedure which was closely analogous to 'assisted spontaneous' repatriation in that transport assistance only was provided and precise numbers could not be calculated in advance. But the numbers sent in any particular convoy can always be controlled through the amount of transport provided and, in the case of Malawi, it would have been relatively easier to hold people back since the great majority came from camps in which they could for a little longer have remained.

(78) These matters have been discussed elsewhere, notably in the Chapter on Malawi, as has much else which concerns the country of origin and its role. The principle of the country of origin as focal point is a sound one and should remain the basis on which future repatriation/reintegration operations are conducted. Without under-estimating the amount of genuine and valuable cross-border consultation that did take place, it could have been more effectively applied - particularly during the mid-1994 phase of 'accelerated' repatriation - during the repatriation from Malawi to Mozambique.

CONCLUSION

(79) UNHCR Mozambique entered the repatriation operation as a modest-sized, under-staffed and under-provided Branch Office with an 'inferiority complex' which it has never entirely shed. After a slow start, it has made enormous strides but would have benefited greatly from a much stronger staffing and logistic base early in 1993 at the latest rather than in late 1993 and 1994. Given the fact that reintegration is in the long run a much more demanding task than the physical process of repatriation, as well as being crucial to the operation's ultimate success, emergency staffing procedures would have been justified at least for Mozambique, if not for the countries of asylum as well.

(80) The principle of the country of origin as the focal point for all UNHCR activity, including the pace of organised repatriation, could have been better observed in terms of coordination with the major country of asylum, Malawi, and should be adhered to in future repatriation operations.

(81) It might be added that the distinction between repatriation and reintegration for evaluation (or any other purposes) is largely an artificial one. Planning for reintegration - the restoration of basic services, Quick Impact Projects (QIPs) and the like - needs to begin before repatriation and, in terms of essential services, implemented before the returnees arrive. Successful reintegration projects act as a 'pull factor' to encourage further repatriation while a lack of reintegration activity can be a powerful deterrent. The ultimate success or failure of a repatriation programme must be judged by the success or failure of reintegration.

PREVIOUS EVALUATIONS

(82) There has been one earlier evaluation of the work of UNHCR Mozambique in the repatriation and reintegration of refugees. This was carried out in November 1994 by ETC International21, on behalf of the Royal Netherlands Government, on the basis of a field visit to Mozambique between 1 and 13 November 1994, including the preparation of the evaluation mission report. The report is highly complimentary of UNHCR and its work in Mozambique but can in no sense be judged a rigorous evaluation of the effectiveness of UNHCR programmes. It advises the Dutch Government to continue its financial support and to "seek ways to enhance financial control, management and reporting capacity in UNHCR to reflect best modern management practice" without, however, any very clear statement of what this practice might be. The report's main concern seems to be what will happen after UNHCR pulls out.

MALAWI

History and Background

(83) In numerical terms, the other five countries of asylum pale into relative insignificance by comparison with Malawi. Crossing the border between Mozambique and Malawi has always taken place and, given the nature of the frontier between the two countries and the close ethnic ties between the people on either side, will certainly continue. But the armed conflict in Mozambique led to a substantial influx from June 1986 onwards, until by September 1987 some 279,000 Mozambicans had sought refuge in Malawi. The influx continued at a varying pace throughout the years that followed, until by late 1992 a peak population figure of 1,058,492 had been reached, the highest single monthly arrival figure of 60,000 being reached in November of that year (drought being a major cause22). But even by 1990, the refugee numbers of over 800,000 amounted to more than 10 percent of the country's population and - at the time - the largest refugee population in Africa and the third largest in the world.

(84) The UNHCR Branch Office at Lilongwe was opened in 1987, followed by a Sub-Office in Blantyre and Field Offices in the main areas of refugee concentration - Lilongwe itself covering Mchinji and Nkhata Bay, Dedza, Ntcheu , Nsanje, Mangochi, Mwanza, Chikwawa and Mulanje. The refugees were either settled among the local population or accommodated in camps spread out between twelve Districts, mainly in the Central and Southern Regions. WFP supplied food to extended delivery points at Lilongwe and Blantyre, while the Malawi Red Cross Society handled the distribution of both food and non-food items in the camps. By early 1993, there were over 130 distribution centres and over 1000 distribution points. In 1993, UNHCR had 22 implementing partners for its care and maintenance programme, including a number of the best known international NGOs but with a number of local NGOs as well. The UNHCR budget reached its peak of US$28 million in 1992 and was still at US$27 million in 1993.

(85) By the early 1990s then, there was an established Branch Office, a well-managed operation and a care and maintenance programme second to none at the time. Excellent relations had been maintained with the Government, which - after many years of deliberate isolation from UNHCR23 - had at last welcomed a UNHCR presence, acceded to the 1951 Convention and 1967 Protocol and enacted its own liberal refugee legislation in the Refugee Act of 1989. A National Joint Operations Committee (JOC) coordinated refugee assistance and the focal point for the refugee programme - initially in the Ministry of Health - was the Commissioner for Disaster Preparedness, Relief and Rehabilitation in the Office of the President and Cabinet.

(86) But the voluntary repatriation of the Mozambican refugees was never off the agenda. There had always been some spontaneous repatriation and a constant movement to and fro of refugees, particularly those from Tete Province who would cross back to Mozambique to cultivate and then return, as the May 1993 Plan of Operation put it (page 26 para.155) "to the relative safety of Malawi where they were part of UNHCR's assistance programme"24. A Tripartite Agreement for voluntary repatriation was signed between UNHCR, the Government of Malawi and the Government of Mozambique on 21 December 1988, over four years before Tripartite Agreements were concluded with any of the other countries of asylum. The agreement was not in the form subsequently adopted as standard but covered all the major issues such as the simplification of border procedures and laid particular emphasis on the preparation, by the Tripartite Repatriation Commission created by the Agreement, of repatriation plans.

(87) Four meetings of the Commission were held between April 1989 and July 1990. At the July 1990 meeting, Malawi and Mozambique jointly presented a draft repatriation plan and a so-called 'final' draft of this document was issued in January 1991 (there had been at least six successive drafts and revisions in the light of comments received from both governments, other UN agencies, the donor community and NGOs). In common with other country plans, it was overtaken by events but the work put into it was far from wasted - the identification of exit/entry points and transit centres within Mozambique was an essential preparatory measure, to give but one example. Nevertheless, a key element of the plan that repatriants would initially move, not to their home destinations but to 'safe areas' or settlement locations in Mozambique (pending final return to home areas) never materialised and one must doubt that it could have been a feasible method of repatriation, except in a very limited way.

(88) The detailed planning for repatriation which took place in Malawi, as elsewhere, well illustrates the Branch Office's commitment to voluntary repatriation as the ultimate durable solution throughout the years when the major emphasis had necessarily to be on care and maintenance operations. There had always been some degree of repatriation activity, inevitable in a context of constant cross-border movement - 8,092 repatriated by UNHCR during 1989 and 1990 and, according to Mozambique Government figures - which must be highly suspect given the context of the times - 200,000 spontaneously returned from neighbouring countries, the vast majority from Malawi. Still it is gratifying that a Canadian CIDA evaluation of UNHCR in August 199225 recorded that:

"UNHCR is confident that the vast majority of refugees in Malawi want to return to their homes and will repatriate as soon as it is safe for them to do so ... the strong conviction that the refugee situation is temporary is a major reason for the perceived (and real) success of the programme in Malawi."

(89) Other studies were not as sanguine. A report prepared for the US Embassy and USAID/Malawi in November 199026 concluded, based mainly on interviews with local leaders, that 'the refugee situation is no longer a temporary one' and 'most Malawians believe that they face a long-term refugee situation'. The authors did not, perhaps, fully appreciate the extent to which Malawians had a vested interest in stressing to aid agencies the horrors of their long-term plight but, prior to the Peace Accord, few would probably have shared UNHCR's confidence in an eventual mass repatriation.

SPONTANEOUS REPATRIATION

(90) In fact, repatriation began only slowly after the signing of the Peace Accord and refugees were still entering Malawi - 60,000 altogether in the months following the Accord, although this was mainly as a consequence of the continuing drought rather than the civil war. Then the pace of repatriation began to gather momentum following a good harvest, until by September 1993 over 250,000 refugees had returned to Tete Province alone, principally from Dedza and Ntcheu. In November 1993, movements increased dramatically until, in the words of Branch Office Malawi27 "with the discontinuation of general food distribution in Dedza and Ntcheu and adequate and regular distribution in the adjoining Mozambican districts of Angonia and Tsangano, refugees returned in droves".

(91) These movements were entirely spontaneous in character - organised repatriation of vulnerable groups did not begin until the end of January 1994. Refugees returned to Mozambique as and when they wished, without registration or Voluntary Repatriation Forms (VRFs) and without surrendering their ration cards before departure. Branch Office Malawi listed among the contributory factors behind spontaneous repatriation the signing of the Peace Accord, the refugees' perception that peace was continuing, a good harvest and availability of food in Mozambique, improved access roads to areas of origin and the effect of the dry season in improving roads and making rivers passable.

(92) Of significance also must have been the discontinuation of food distribution referred to above, which began in Dedza and Ntcheu at the end of November 1993, when it was considered that three-quarters or more of the refugees had already returned to Mozambique, and was progressively extended to other areas thereafter. The euphemism adopted to describe this procedure was 'food consolidation' but it was consolidation only in the sense that the distribution of basic rations was restricted to a smaller residual population and to fewer and fewer areas as the great majority of refugees in particular settlements or camps were seen to have gone back to Mozambique. In other circumstances, the cutting-off of rations has been rightly considered a coercive measure introduced (usually by inhospitable host governments) to compel refugees to return to their country of origin and thus the antithesis of everything UNHCR stands for in safeguarding the principle of repatriation as a voluntary act.

(93) In the Malawi context, however, with many thousands of refugees living in areas immediately adjacent to the Mozambican border - a substantial section of the main trunk road from Lilongwe towards Zomba and Blantyre runs along the border in Dedza and Ntcheu districts - food consolidation became inevitable. Many refugees who had already repatriated to Mozambique continued to return to Malawi to collect their rations at each fortnightly distribution and would doubtless have continued to do so indefinitely had they been permitted to do so28. The discontinuation of food supplies in Malawi once the preponderance of refugees had repatriated, was essential if this cross-border activity was not to continue forever and if the residual population were to be persuaded (not forced) to 'take the plunge'.

ORGANISED AND ASSISTED SPONTANEOUS REPATRIATION

(94) A particular focus of tension between the UNHCR Branch Office in Malawi and the Branch Office in Mozambique has been the practice, initiated by BO Malawi, of 'assisted spontaneous repatriation'. The Branch Office has explained29 that, as spontaneous repatriation gained momentum in late 1993 and early 1994, it was noticed that vulnerable categories of refugees and those whose places of origin in Mozambique were not easily accessible were being left behind. From this unsatisfactory situation was born, in the words of the Branch Office, 'the idea of organised/assisted repatriation for these groups', an idea which was endorsed at the 8th meeting of the Tripartite Commission on 28/29 March 1994.

(95) In fact, the concept of organised repatriation had clearly been present from the earliest days of planning for voluntary repatriation in the late 1980s and had already started in other countries of asylum, beginning with the first Zimbabwe convoys in June 1993 and followed shortly after by organised movements from Swaziland and Zambia. There was nothing controversial or contentious about this and, except in Malawi, it was the only form of repatriation with which UNHCR was directly associated. In Malawi and South Africa, organised repatriation involved principally the movement by buses, trucks or rail of vulnerable cases and others who were unable to travel unassisted; the first convoy of 41 individuals took place from Dedza on 31 January 1994. In addition to movements by road or rail 2,527 refugees were transported by ship across Lake Malawi from Nkhata Bay in April and May 1994. The majority of movements, however, were either (I) by road from camps in Mwanza District to Nsanje, an average distance of 275 km and then by bus to Mutarara in Mozambique or by boat across the Shire River to Megaza transit centre en route to Morrumbala or (ii) by rail from the area of Nyamithuthu camp to Nsanje and thence to the border by bus or by boat to Megaza.

(96) BO Malawi statistics record that a total of 128,340 refugees had returned to Mozambique under organised/assisted repatriation arrangements up to the end of February 199530. There is no breakdown in these figures between organised repatriation on the one hand and 'assisted spontaneous' repatriation on the other and the distinction between the two is not always easy to draw. Nevertheless, the 'assisted spontaneous' movements became particularly controversial at certain points in time and sources in BO Malawi have described them to the evaluation mission as embodying a concept which their counterparts in Mozambique either did not, or refused to, understand. One response from Mozambique characterised assisted spontaneous repatriation as 'dumping at the border' a reaction which seems also to have been evoked by some of the 'organised repatriation' movements by road or rail.

(97) If we define assisted spontaneous repatriation as the provision of facilities aimed at helping those who are moving of their own accord to do so in dignity and safety, then a prime example is the so-called 'Lower Shire Boat Repatriation Exercise' which began operations in July 1994. Until that time, refugees returning to Morrumbala District in Mozambique crossed the Shire River by means of traditional dug-out canoes, which could safely carry only a few passengers and limited baggage and ran the risk of capsizing if over-loaded. The alternative route by road was 1,200 km in length. Under implementing arrangements between UNHCR and the American Rescue Committee (ARC) and with funding support from October 1994 onwards from the US Department of Defense, five locally built boats were provided to operate from crossing points at Nsanje, Tengani and Chiromo. The Nsanje crossing involved a three-hour trip down river to Megaza in Mozambique and a four hour return, making seven hours in all. From Tengani and Chiromo, simple river crossings of a few minutes only were involved. Each boat could carry up to 50 passengers and fairly substantial quantities of luggage, including bicycles and furniture.

(98) Most of the early boat crossings fell in the category of organised rather than assisted spontaneous repatriation in the sense that refugees were taken, for example, by bus to Nsanje from the Mwanza camps and thence by boat to Megaza transit centre in Mozambique for onward transportation to their final destination, often via another transit centre in Morrumbala town 45 kms further distant from Megeza. They were registered for departure and their ration cards withdrawn prior to leaving their camp in Malawi.

(99) By 6 October 1994 - prior to a brief suspension of the programme following an outbreak of bubonic plague in Mozambique - 7,642 returnees had crossed by boat to Mozambique through the UNHCR/ARC operation, of whom 5,146 had gone via Nsanje and the remainder via Chiromo or Tengani.

(100) By late November 1994, however, and continuing through the early months of 1995, difficulties and differences arose. These were part of a wider picture, with UNHCR Mozambique complaining that many more refugees were being repatriated than had originally been agreed31, objecting to erratic movement patterns and repatriation during the rainy season when roads in Mozambique were impassable and expressing grave concern at a 25-26 February 1995 distribution of 3-months rations to the residual caseload in Nsanje district which, it feared, might precipitate a mass movement to Mozambique just when conditions for the reception of repatriants were at their most precarious32. A good deal of heat and cable traffic was generated, with UNHCR Malawi insisting, inter alia, that it had never contemplated large scale repatriation during the rainy season and that the justification for a longer food distribution period was, as on earlier occasions, to discourage those who had already repatriated to Mozambique from returning to Malawi on distribution days to try and collect the rations to which they were no longer properly entitled. It emphatically did not imply that Malawi was expelling the remaining refugees to Mozambique.

(101) The significance of the boat exercise in this context is that, at least in its latter stages, it represents in microcosm the issues on which opinions diverged between BO Malawi and BO Mozambique. Feeling so strongly that refugees should be dissuaded from returning until May 1995 when the rains would be over, UNHCR colleagues in Mozambique found it difficult to understand in February and March that the crossing point at Nsanje remained open with three boats regularly available for the use of refugees wishing to cross the Shire River to Megaza. Given its attitude that, to quote a UNHCR Maputo cable of 24 February 1995, "it is unwise to repatriate refugees during the rainy season ... road travel inside Mozambique is difficult and often impossible ... it is onerous and difficult for refugees to reestablish their houses ... the rainy season is the time of high malnutrition rates.... ". BO Mozambique clearly felt that assisted movements should be suspended and the boats taken out of service.

(102) BO Malawi, on the other hand, always insisted on the spontaneous nature of the repatriation movements and on the likelihood that repatriating refugees were perfectly well aware of conditions on the other side of the river and would cross when they wanted to whether UNHCR - sponsored boats were or were not provided. Certainly, when the evaluation mission visited the Nsanje crossing point on 18 March 1995, boats could be observed crossing in both directions and one old man interviewed briefly on arrival from Mozambique said that conditions there were 'good'. The boats continued to operate during the rainy season and by mid-March some 15,000 in all had been repatriated by this means. By the end of March 1995, UNHCR Mozambique estimated that all the refugees expected to arrive via Megaza had already done so, while 95% of the returnee caseload to Morrumbala District was believed to have returned in 1994.

(103) Organised repatriation movements by truck or rail had virtually come to an end by December 1994 - the IOM internal transport operation via Mutarara in Mozambique came to a halt on 17 December when the rains made roads impassable and was not to be resumed, if at all, until mid-May 1995. According to UNHCR Mozambique, this so-called 'Mutarara Special Operation' had assisted 93,613 refugees to return home by the end of 199433 by transporting them from the Villa Nova entry point bordering Nsanje district to their various destinations in Mozambique.

(104) Organised (or was it assisted spontaneous?) repatriation by road also brought periodic tensions and misunderstandings. For example, on 7 May 1994 the Field Officer, Milange in Mozambique complained that assisted arrivals totalling 2,867 persons during the previous week had exceeded by 100% the target figure communicated to his counterpart in Malawi in pre-movement consultations and that local transportation and reception facilities were severely strained. Some 6,000 spontaneous arrivals had taken place over the same period, adding to the problem of absorption. There were other complaints concerning overloaded buses and excessive numbers arriving on a single day. By 14 May, however, Field Officer Milange was able to report that the situation was under control and that coordination between Field Offices on both sides of the border was improving.

(105) It is important to put these and similar problems into a true perspective. May 1994 was a peak movement for repatriation, and handling capacity on all sides had reached its limits - Milanje and Northern Morrumbala were receiving an average of more than 1,000 returnees each day and between early May and early June over 138,000 refugees returned spontaneously or with assistance from Malawi to Mozambique. Overall, the great majority of the 1.06 million refugees who were in Malawi when the refugee population was at its very peak in February 1993 repatriated spontaneously without assistance of any kind and most organised or 'assisted' repatriation took place smoothly and with full cooperation on both sides. Perhaps the underlying cause of tension and mistrust between the two Branch Offices was a feeling on the part of UNHCR Mozambique that the pace of repatriation (particularly when it seemed to have been deliberately accelerated by food consolidation and the provision of additional transport for assisted spontaneous return in the middle months of 1994) was being dictated by the country of asylum rather than by the country of origin of the refugees.

(106) In a sense, as one senior desk officer has suggested, both sides were right. As a matter of principle, the pace and logistics of repatriation must be determined by the absorptive capacity of the country of origin and this principle was carefully observed in all the other countries of asylum. On the other hand, numbers elsewhere were relatively insignificant compared to those in Malawi and, except in Zambia, the refugees for whose repatriation UNHCR assumed direct responsibility were generally accommodated in camps or settlements far from the Mozambique border. Only in Malawi did the great mass of refugees have the option of making their own way home, often to areas with which they had maintained close contacts during their years as refugees. In short, the refugees repatriated themselves and UNHCR's role was mainly to ensure through, for example, food consolidation that, having gone, they stayed. There is no doubt, however, that by early 1994 the emphasis, from UNHCR Malawi's point of view, had changed from facilitating repatriation to actively promoting and, indeed, accelerating it. At an extraordinary coordination meeting on 23 May 1994, the UNHCR Representative for Malawi "emphasised the importance to seize this opportunity to accelerate the repatriation process". He cautioned all concerned not to confuse assisted spontaneous with organised repatriation and to act accordingly. Nevertheless, as has been remarked earlier, the distinction between the two is not easy to draw but seems to lie, if anywhere, in the relatively unregulated and unpredictable size and frequency of assisted spontaneous movement. This in turn created the reception and absorption difficulties of which UNHCR field staff in Mozambique complained and which they believed could easily have been avoided had their counterparts in Malawi not been under pressure to "repatriate at any price".

(107) For all the conflicting pressures in the field, the mechanisms established for coordination between UNHCR Branch Offices, the two governments and operational partners on both sides of the border were at least as well developed between Mozambique and Malawi as with any other country of asylum. The Tripartite Commission held eight meetings between April 1989 and March 1994 (although it seems to have lapsed thereafter and there was a gap of over a year between the sixth and seventh meetings). NGO representatives were invited to attend as observers from the fifth meeting onwards. Cross-border district coordinating committees met sporadically and National Coordinating Committees were established in Malawi in June 199034 and in Mozambique in July 1990. Most importantly, UNHCR Malawi/WFP/UNHCR Mozambique held frequent coordination meetings, eight in all up to the end of March 1995, supported on issues of detail by a joint Malawi/Mozambique Repatriation Task Force during late 1993 and the first half of 1994.

(108) The deliberations and conclusions of all these bodies are fully recorded. While they do not, by and large, reveal the frustrations and resentment that occasionally erupted, they do demonstrate that overall a high degree of coordination, cooperation and mutual understanding was achieved. The basic lesson to be learnt, was perhaps best expressed in the following words in a report by the Field Office, Lilongwe on the organised repatriation to Niassa Province from Nkhata Bay:

"Another learning experience of the repatriation is the importance of careful planning and constant review of the progress in an organised repatriation. The operation also showed the importance of working as a team in that there were a number of agents with different specific tasks but all aimed at one purpose and goal".

(109) The Nkhata Bay operation successfully moved an entire caseload35 of 2,492 refugees back to Lago District in Mozambique by ship between 30 April and 13 May 1994. The repatriation plan had been approved by the National Coordination Committee in February and pre-departure activities had begun in March. A number of NGOs, particularly the Norwegian Refugee Council, were fully involved and there was close cooperation with the Government authorities on both sides. The exercise appears to have been a model of its kind and the description in the final report of the various planning and operational phases which led to its success should be required reading for those planning future operations of a similar kind.

(110) Assisted repatriation from Malawi to Mozambique finally came to an end with the so-called 'final assisted movement' from the remaining Nsanje District camps to Mutarara and thence to final destinations in the Tete, Sofala and Zambezia Provinces between 22 May and 13 June 1995. Over this period, a total of 41,115 returnees were moved by UNHCR and - in Mozambique - by NAR, using directly hired transport in the form of buses, trucks and two boats for the Zambezi river crossing, mainly for those whose final destination was in Sofala Province. There was no IOM involvement and, despite heavy rain, periodic overcrowding with up to 2,000 inhabitants at the Bauer Transit Camp in Mutarara and the fact that two boats proved insufficient (one was frequently out of order), the operation appears to have been a great success, with excellent cooperation on both sides. A peak figure of 3,349 was moved from Malawi to Mozambique on a single day - 7 June. Happily, in the words of the Field Officer, Mutarara, repatriants were arriving "with huge stacks of food".

INFORMATION

(111) The need for refugees in Malawi to be properly informed about all aspects of the voluntary repatriation process and, in particular, about conditions in Mozambique was constantly stressed. The minutes of Tripartite Commission meetings from 1989 onwards contain many references to 'the need to publicise the possibilities for repatriation' (April 1989). Between 27 December 1989 and 10 January 1990, the Malawi Minister of Local Government together with representatives of the Mozambique Government and UNHCR visited all the refugee camps in Mulanje, Mangochi, Ntcheu and Dedza districts 'to acquaint refugees' with the purpose of the Tripartite Commission and to sensitise them to the possibility of returning to Mozambique in safety and dignity. The need for high-level delegations from both countries and UNHCR was again emphasised by the Tripartite Commission's working group in July 1993 and a 1994 report pointed out that "virtually every planning document written over the last two years concerning facilitation of voluntary repatriation from Malawi contains some reference to the need for a mass information campaign"36.

(112) Visits by Tripartite Commission members to refugee camps and areas, although regularly described as an important information tool, seem after the beginning of 1990 to have taken place only infrequently and then by relatively low-level delegations. In a meeting with the Mozambique Embassy in Malawi, the evaluation mission was told that there were too few visits by Mozambican delegates to the refugees. The first official published notification to the refugee community that 'those who wish can now go home to Mozambique' was only signed on behalf of the Tripartite Commission on 29 March 1994 and issued in April 1994, rather late one might think with little more than half of the peak refugee population then remaining in Malawi. By that time also, a substantial degree of food consolidation had already taken place and there seems no doubt that this was generally taken by the refugees to mean that it was time for them to leave37.

The Tripartite Commission message was read over the radio in Malawi in different languages38 and distributed in the camps as posters. It was supplemented by an informal refugee information network initiated by SCF (UK) in collaboration with UNHCR between January and July 1994. This was designed to disseminate information collected by refugees themselves on conditions in Mozambique from those who had crossed the border and returned to Malawi. While needing to be applied with caution, the network had obvious merits, not least the ability to provide area-specific information as a general information pamphlet could not do.

(113) The lack of area-specific coverage was one of many criticisms of UNHCR's information activity made by the SCF (UK) report referred to above. The report criticized the fact that the 29 March 1994 message from the Tripartite Commission was 'late in the extreme' and added that the broad nature of the information provided meant that it was not always accurate - citing a report from the UNHCR Field Officer in Mangochi that refugees who had returned to Mozambique 'particularly after the information packs were distributed in April 1994' were coming back to Mangochi for lack of food. More seriously, the report suggested that the issue of the information bulletin signalled 'a change from unbiased information towards low-key propaganda.' It is certainly true that, by early 1994, UNHCR representatives at joint meetings with the Government were giving assurances that, for example, while the voluntary nature of their decisions would be respected, the repatriation message would be strongly conveyed to refugees.

(114) There were other SCF (UK) criticisms and generally the report is fairly jaundiced in its attitude to UNHCR and officialdom generally. A more temperate view might be to agree that the bulletin was certainly late in being issued, given that so much repatriation both from Malawi and other countries of asylum39 had already place but that SCF (UK) probably underestimated the extent and accuracy of the information already available to the refugees via their own channels and cross-border movement and promoted by the SCF (UK)-sponsored informal information network. Nevertheless, bearing in mind the statement in UNHCR's own January 1994 Progress Report (page 14 para.81) that "much more attention will have to be paid this year to a good mass information campaign to ensure that refugees are aware of developments in Mozambique", more might have been done and earlier to disseminate information on particular areas in Mozambique and on the situation vis-à-vis food availability, roads, schools, health services, water supply and mines. This presupposes, however, the availability of accurate information from UNHCR in Mozambique which, until it strengthened its presence there in late 1993 and early 1994 (mainly the latter), was far from being the case. Only the Milanje Field Office was opened in 1993 (September) and Mandimba and Morrumbala not until March/April 1994.

(115) Finally, it should be mentioned that the information needs of international agencies, NGOs (although these often had their own information packages), diplomatic missions, government departments and the like were met by a Branch Office Malawi publication, "Zikomo Malawi"40 of which there were five issues covering the "boom"repatriation period between 9 May 1994 and 30 September 1994. Unlike the information bulletin, which was distributed to the refugees in Chichewa, Chisenna and Portuguese as well as English, 'Zikomo Malawi' appeared in English only and, while giving useful publicity to the repatriation process, was not intended as an information vehicle for refugees or circulated within the refugee camps.

MINE AWARENESS

(116) The fact that the bulk of the repatriation from Malawi was spontaneous did not prevent a mine awareness campaign being mounted among the refugees, on similar lines to campaigns in other countries of asylum. In March 1993, before the major wave of repatriation was under way, UNHCR approached the International Rescue Committee (IRC) to become its implementing partner in a Land Mine Awareness Programme (LMAP), an appropriate choice since IRC had previously been involved in mine awareness campaigns in Thailand and Pakistan and had played a major role in refugee assistance in Malawi since 1987. Concurrently, UNHCR secured the support of Norwegian People's Aid (NPA), the NGO primarily responsible for demining in Mozambique itself, as consultants.

(117) The programme required two separate approaches: formal instruction in the camps and an informal mass-information campaign in those areas where the refugees were integrated with the local Malawian population. Between September 1993 and June 1994, the LMAP trained a total of 180 instructors and 28 supervisors who worked in 17 refugee camps in the Southern region of Malawi training an estimated 380,000 refugees. According to IRC, however, substantial funding was never secured for a mass information campaign in the central districts where most of the refugees who had integrated in Malawian villages were located. Apart from the display of posters and the distribution of other materials, little could be done before the massive waves of repatriation from those districts to Mozambique took place. Fortunately, the areas mainly in Tete Province to which they returned were considered to have been relatively lightly mined. IRC's own records of mine victims show reports of only 30 cases from 1987 onwards including 8 deaths, although there is no suggestion that these figures are exhaustive.

(118) In summary, one has to conclude that probably less than half of the refugees repatriating from Malawi had much systematic exposure to the mine awareness campaign and that only those in camps were covered in a comprehensive way. On the other hand, refugees repatriating spontaneously from other countries of asylum - Tanzania, Zimbabwe - had virtually no mine awareness instruction and the total number covered by the Malawi programme was much greater than elsewhere.

ENVIRONMENT AND INFRASTRUCTURE

Environment

(119) In every country with a sizeable refugee population, UNHCR must be concerned with the impact on the environment and, once the refugees repatriate, with the disposal of the physical assets associated with the care and maintenance programme instituted to assist them.

(120) In other countries of asylum hosting Mozambican refugees, the environmental impact was limited by the establishment of camps or agricultural settlements and, of course, by the much smaller number of refugees. Similarly, the disposal of assets - although conscientiously attended to in each case41 - did not constitute a major problem. In Malawi, however, the environmental and ecological consequences of the presence of over a million refugees, many of them living in so-called 'open settlements' alongside Malawian villages or in open camps, were enormous. A complex physical infrastructure had, moreover, been put in place to provide for refugee needs and represented a major asset either to be used for the national benefit or left to waste.

(121) To its credit, UNHCR Branch Office in Malawi recognised the environmental problem from the start. The rate of deforestation - 150,000 hectares or 3.5% of the national forest cover - is the highest in southern Africa with the great majority of people deriving their total energy needs from wood. This rate was undoubtedly aggravated by the presence of the refugees who cut most of their fuel requirements from the scarce wood resources of the areas where they lived, as well as using forest products for the construction of shelter and other domestic needs. UNHCR therefore launched a reafforestation programme in 1988 and by mid-1994 had spent some $2.8 million on the planting of 8146.84 hectares of forest in refugee-hosting districts. The German Government has also provided substantial funds through UNHCR for the development and conservation of national resources in refugee areas. Project activities have also included erosion control, the production of fuel saving stoves and training.

(122) This effort, however commendable, has been only 'a drop in the ocean'. At its peak, the refugee population was denuding the forest cover at the rate of 22,000 hectares per year while UNHCR was replacing it at an average rate of 1,300 hectares per year - the shortfall is only too visible in the vicinity of any of the major refugee hosting sites42. The Branch Office has been doing its utmost to interest other potential donors, such as the EC and UNDP, in assisting the Government of Malawi in incorporating forestry and other needs into their country programmes43 and regards this as an integral part of its approach to repatriation and the achievement of a durable solution to the refugee problem. This is an important lesson for any voluntary repatriation operation; the range of activity undertaken must extend far beyond the mere physical process of repatriation to the restoration of the environment in the country of asylum as well as the successful reintegration of the refugees in their country of origin.

(123) One more example - others could be given - of the environmental impact of refugees in Malawi is the damage sustained to the road system by the constant transportation of relief supplies, particularly on secondary and district roads which were not designed or constructed to carry heavy traffic. The volume of relief goods being carried an average of at least 50 kilometres from reception point to camp distribution centres, reached more than 180,000 tons annually, causing severe damage to road surfaces, bridges and culverts affecting not only access to refugee centres but the normal transportation of agricultural produce from Malawian farms. UNHCR has maintained a total of 1,073 kilometres of road in recent years and, although this has been phased down with the departure of the refugees, maintenance has continued in Nsanje District in 1995.

Infrastructure

(124) After food and transportation, the highest single item in the Malawi care and maintenance budget was health - implemented at its peak by the Government and six NGOs, of which only one, MSF, was left in Nsanje by March 1995. But 800 Ministry of Health staff were still employed on the refugee budget, and country-wide there were a total of 19 hospital wards, 91 staff houses, 17 dispensaries, 14 maternity units and 36 other structures which had been built under UNHCR auspices to meet refugee needs and were to be handed over to the Government. In education, there were 437 classrooms and furniture to be handed over to the Government, which hoped to use these school facilities for the national education programme. A total of 1,428 water points of various kinds had either been capped or handed over to the Malawi Water Department.

(125) Great care has been taken by Branch Office Malawi over the phase-out modalities involving these and other assets, as well as many project vehicles and large numbers of supporting staff. Hand-over plans to government ministries have been prepared and discussions initiated with other international agencies, not only UNDP and the EC, but also UNICEF - particularly as concerns the water and health sector - and others. There is no need to elaborate here (and a formal handover to the Government has already taken place) except to reiterate the importance of an orderly phasing-out of relief assistance and the disposal of assets to the best advantage of the host nation which has borne the burden of the refugee presence. UNHCR Malawi has rightly discharged these responsibilities as an integral part of the repatriation process.

(126) One final point - it is extremely important that phase-out arrangements and negotiations be conducted with great sensitivity and diplomacy, particularly where the host government is concerned. The departure of the refugees means loss as well as gain to the host society and the need to assume responsibilities and administer assets which have hitherto been handled by external agencies may come as both a challenge and a shock. There is a natural desire to 'squeeze the lemon' for every last drop of aid before it is too late. Future good relations may well depend on the spirit and manner in which the phase-out process is carried out.

(127) The Government of Malawi has remained extremely anxious to secure continuing support for the restoration of the environment and the maintenance of the physical assets which are now its responsibility to administer. These hopes and expectations are far beyond UNHCR's capacity to realise and must be left in the long term for development agencies to meet under the so-called transition from refugee aid to development - usually effected with disappointing results. Any evaluation must conclude, however, that UNHCR has been alive to the issues and has done all it could. Indeed, it was largely in the spirit of offering some infrastructural support that UNHCR undertook the rehabilitation in 1994 of a section of railway line to transport refugees under 'assisted spontaneous' arrangements between Nyamithithu and Markha on the Mozambique border - a purchase of goodwill at relatively little cost44, although it seems doubtful whether much more of the railway link will be made operational in the foreseeable future.

CONCLUSION

(128) Despite the burden of a caseload of over a million Mozambican refugees, UNHCR in Malawi had distinct advantages in the task of initially facilitating and then promoting their voluntary repatriation. There was already - quite unlike the situation in Mozambique - a developed network of Sub and Field Offices, a strong NGO presence, a cooperative government and an orderly and law-abiding community of refugees. Care and maintenance facilities had been developed to an impressive standard.

(129) There were no serious protection problems in Malawi. The government machinery was generally honest and efficient and, although it had made nine reservations, Malawi was faithful to the letter and spirit of the 1951 Convention, the 1967 Protocol and the OAU Convention. There was no hostility towards the refugees, most of whom shared an ethnic heritage with their Malawian cousins. A repatriation training workshop originally planned for July/August 1993 for UNHCR staff, government officials and NGOs was never held, presumably for the simple reason that it was not needed.

(130) Within the Branch Office, a Senior Repatriation Officer arrived in August 1993 to support the Representative in directing the repatriation programme. A total of 7 International and 5 National Officer posts for the repatriation operation in Malawi were approved in May 1993 with supporting GL staff (mostly in telecommunications). The lack of protection problems (see above) meant , however, that one P-3 Protection Officer post for Blantyre never materialised while the other Protection Officer post was used mainly for the Nkhata Bay repatriation and the urban caseload. The reinforcement of repatriation capacity was further enhanced by the simultaneous appointment by WFP of an Emergency Officer to deal with repatriation issues. Later WFP established a regional project to allow flexibility in food allocations between Malawi and Mozambique, which proved invaluable to the introduction of food consolidation by making it reasonably possible to ensure that food distribution was only stopped on one side of the border when it was regularly available on the other. Teamwork between agencies was excellent. Joint UNHCR/WFP Malawi/Mozambique meetings were regularly held; the evaluation mission has the record of eight such meetings between late 1993 and March 1995. Cooperation between the two agencies most intimately concerned with voluntary repatriation was therefore constant and close.

(131) In the view of some participants in the process, mid-1993 marked a transition in the role of Branch Office Malawi from facilitating voluntary repatriation to actively promoting it, influenced no doubt by the Deputy High Commissioner's mission to Mozambique between 6-12 June 1993 after which he urged 'a complete and rapid transformation of our programmes in the asylum countries and in Mozambique into a fully-fledged repatriation operation'. (DHC's Mission Report 18 June 1993 page 3 para 12.) Most significant of all, however, must have been the example of spontaneous repatriation that refugees returning to the Angonia and Tsangano Districts of Tete Province had already set, with lesser numbers returning to Macanga and elsewhere. In June and July 1993, over 70,000 refugees returned to Angonia District alone.

(132) It was therefore - and this is the principal (and obvious) lesson from the Malawi experience - the refugees themselves who played the predominant role in repatriation from Malawi and who made the largest single contribution to the operation's success. This is agreed by all involved and to make the point is not in any way to belittle the excellent work of the Branch Office in facilitating, promoting, organising and assisting the repatriation process. For example, an admirably concise and precise repatriation plan was issued by the Branch Office's Repatriation Unit on 19 April 1994 covering all forthcoming organised and assisted movements of vulnerable and other cases, including the operation from Nkhata Bay. But it serves as a reminder to all of us that refugees are human beings with as great a capacity to make up their own minds as anybody else and will not wait for UNHCR or any other agency to arrange for them things which they have already decided to undertake themselves.

(133) The two most controversial aspects of the operation were (I) food consolidation and (ii) assisted spontaneous repatriation.

(134) Food consolidation was defined in the course of a joint UNHCR/WFP food assessment in the Angonia/Tsangano areas of Mozambique and adjacent to Dedza and Ntcheu between 20-22 September 1993 as "food distribution ceases on the Malawi side in a context of an assured and regular food distribution on the corresponding Mozambique side".

(135) The ungainly acronym coined to describe the concept was EFMCM or "Ensure Food in Mozambique and Consolidate in Malawi". Had there been no food consolidation, there seems no doubt that a 'revolving door' syndrome would have been perpetuated - as, indeed, had already begun - and that the Government of Malawi would have insisted on the cutting off of rations had UNHCR/WFP on both sides of the border not taken the initiative themselves. An inevitable policy, then, although one which was open to misinterpretation and which, in some eyes, raised questions about the truly voluntary nature of the repatriation process. It was certainly a message to the refugees, if they needed it, that it was time to go home and no clear arrangements appear to have been made to meet the case of refugees who did not wish to repatriate and elected to remain in Malawi instead. One has also to say that 'an assured and regular food distribution on the Mozambique side' does not always seem to have been provided as a result of poor road communications and the like, so that the strict criteria for the policy were not consistently met. Finally, the most recent consolidation decision of all in Nsanje District, when the 25-26 February distribution of a 110-day ration was announced as the very last, was the subject of an initial serious misunderstanding between the two Branch Offices. The situation was clarified at a joint UNHCR/WFP Malawi/Mozambique meeting on 27 March 1995 but better communication could probably have avoided misapprehension from the start.

(136) 'Assisted spontaneous repatriation' was also a cause of tension, Branch Office Malawi regarding it as distinct from the organised repatriation of (mainly) vulnerable groups. Only the vulnerable cases were repatriated on the basis of Voluntary Repatriation Form (VRFs), while assisted spontaneous repatriation was regarded as merely the provision of transport by road, rail or boat to help refugees do what they would otherwise have done, but with more difficulty, on their own. To Branch Office Mozambique the distinction was specious - UNHCR Field Offices in Mozambique, it believed, had a responsibility to manage the reception and onward movement of 'assisted spontaneous' returnees in just the same way as that of organised groups.

(137) This divergence of opinion was clearly illustrated in the minutes of a meeting between Sub-Office Tete and Branch Office/Sub Office Malawi representatives at Blantyre on 9 May 1994 to discuss the need to suspend assisted spontaneous repatriation to Mutarara District following the discovery of land mines on the Mutarara-Villa Nova road. To Sub-Office Tete, the need for suspension was axiomatic but, while UNHCR Malawi concurred, it obviously had misgivings as the following extract from the minutes of the meeting shows:

'The Malawi side, while appreciating the above-mentioned statements, made the following observations:

The assisted spontaneous repatriation was designed to facilitate and accelerate the voluntary return of those refugees who were relocated by UNHCR from their points of entry to camps further inside Malawi. These refugees have repeatedly expressed their wish to return to Mozambique should transport be made available. The Tripartite Commission at its 8th meeting endorsed the proposal and decided that bus shuttles, railway services and boats at Shire and Ruo crossing points should be provided on an ad hoc and flexible basis. The whole assisted spontaneous repatriation operation was not intended to be an organized repatriation.

The acceleration of voluntary repatriation has been viewed as vital to the timely return of Mozambican refugees to their country in order to register and participate in the forthcoming election and the reconstruction of Mozambique.

That the general announcement to refugees regarding the conditions in Mozambique has generated an immediate positive response for return. Branch Office, Malawi has been under tremendous pressure from refugees to accelerate assisted spontaneous repatriation.

The key to the success of this operation was its flexibility and the ability to move the returnees further inland enabling them to reach their villages of origin at the shortest time possible.

The abrupt suspension of the Mutarara-bound shuttles created problems in Malawi, particularly for those refugees who had demolished their dwellings in readiness for departure. The congestion precipitated by the suspension created potential health and sanitation hazards in temporary shelters.

It is, however, hoped that the demining of the Charre-Villa Nova stretch will be expedited to enable Branch Office, Malawi to honour its contractual obligations with transporters".

(138) The irritation evident in the Malawi side's response was clearly exacerbated by the fact that Sub-Office Tete also asked that, when assisted spontaneous repatriation resumed via Villa Nova, it should initially be limited to 500 individuals per day and that to northern Morrumbala District it should be put on hold because the road network was non-existent and most areas were inaccessible. The conclusions of the meeting did not reflect agreement to either of these requests.

(139) One need not elaborate further or exaggerate the problems. As one senior UNHCR official in Malawi pointed out to the evaluation mission (and as we have noted elsewhere) the respective Branch Offices had different, even conflicting, agendas so that some disagreement was inevitable - Malawi's perceived task was to facilitate, promote and, at times undoubtedly, to accelerate voluntary repatriation, Mozambique's to receive, assist and reintegrate returning refugees. In the last analysis, however, and while appreciating the immense pressures which UNHCR Malawi had to face, one must reiterate the principle that it is not the country of asylum but the country of origin that must dictate the pace of any repatriation which UNHCR has the power to regulate even if only to a limited degree, and that clearly applies to 'assisted spontaneous' as well as to 'organised' repatriation.

SOUTH AFRICA

HISTORY AND BACKGROUND

(140) UNHCR first established a presence in South Africa with the signing of a Memorandum of Understanding with the South African Government on 4 September 1991. The objective at that time was to assist with the return of South African exiles from other countries and to help with their reintegration through training in vocational skills. The Memorandum of Understanding did not give UNHCR access to refugees inside South Africa and this was only achieved with the conclusion of a Basic Agreement with the Government two years later on 6 September 1993. This was quickly followed by a Tripartite Agreement in the usual terms between the Governments of South Africa and Mozambique which was signed in Maputo on 15 October 1993.

(141) Progress was rapid because both governments were agreed with UNHCR on the need for a repatriation programme and were anxious to cooperate in its inception. The Senior Protection Officer from the UNHCR Office in Johannesburg and the responsible Deputy Director in the South African Ministry of Home Affairs had, in fact, visited Mozambique in July 1993 to negotiate the Tripartite Agreement and its basis was quickly put in place. There was, in short, no difficulty in securing the legal framework for repatriation to begin.

(142) The problems were, nevertheless, enormous and, in some crucial aspects, more formidable than those encountered in any other country of asylum. UNHCR had arrived late in South Africa, because of obvious political constraints, and had to forge new relationships with governmental authorities to which, until the demise of apartheid, the very name 'United Nations' had been anathema. It had also to develop partnerships with NGOs which had, in the past, usually been hostile to the government and suspicious of organisations which cooperated with it. It had no clear idea of the potential caseload of Mozambican refugees seeking repatriation from South Africa and with many thousands of Mozambican economic migrants in the country, some working legally in the mines and elsewhere, others illegally as farm labourers, had to adopt a 'rule of thumb' approach as to who was a refugee and who was not.

(143) The uncertainties of the situation had already obliged UNHCR to make clear in the Plan of Operation and Appeal of May 1993 that no actual planning of repatriation from South Africa had been carried out and to ascribe only a provisional figure of US$12 million to this element of the budget. The Plan of Operation45 quotes an estimate by the South African Council of Churches46 of 250,000 refugees in the country and refers to attempts to compile data from the records of church and other humanitarian organisations. The Executive Summary of the Plan foresees an assessment mission to South Africa in May/June 199347, but this did not take place presumably because no basic agreement had yet been reached. However, UNHCR had provided funding for a survey of 6,348 Mozambican households in South Africa which was carried out by the Bureau of Refugees of the South African Catholic Bishops Conference (SACBC) as the so-called "Masungulo Project' between March and August 1993 (report issued October 1993). Rather frustratingly, the survey says nothing about total refugee numbers, except to repeat in the Preface the figure of 250,000 Mozambican refugees in South Africa, on which Catholic and other Christian denominations seem to have been agreed.

(144) The SACBC survey was, nevertheless, able to conclude, inter alia, from the households surveyed, that:

(1) 90.9% of refugees had left Mozambique because of the war and only 6.7% for economic reasons;

(2) 79% of refugee households had a family in Mozambique;

(3) 90.1% had arrived in South Africa between 1985 and 1992 and only 1.5% before 1980;

(4) 83.7% of households wanted to return to Mozambique, the great majority to Gaza or Maputo Provinces.

(145) The survey report is not a particularly impressive document. How does one define, for example, in African extended family kinship terms, having a family in Mozambique? Nevertheless, it does conclude that, despite admitted problems - 'the quality of data collected was found good enough to answer the primary questions about their basic socio-demographic profile and intention to or not to return to Mozambique'. It should also be borne in mind, however, that the survey - based as it was on a total of 6,348 households with an average of 4.4 persons per household - could only represent at best the intentions of a minority sample of the refugee population.

(146) Meanwhile, the South African authorities were regularly deporting illegal immigrants from Mozambique, a process which continues unabated to this day.48 The South African Defence Force had also been helping to repatriate Mozambicans with transport and relief as they returned home spontaneously through the Kruger National Park, and between November 1992 and December 1993 some 10,000 persons were assisted in this way. Both the South African Government and UNHCR were anxious to phase out this operation in favour of a more regular UNHCR-based approach. By the end of November 1993, therefore, the UNHCR office in South Africa had conducted its own preparatory protection survey to determine the whereabouts of Mozambican refugees who were distributed mainly in settlements or scattered groups in the Northern and Eastern Transvaal, and had finalized an operational plan in the form of recommendations by an Inter-Agency Needs Assessment mission in late November in which representatives of both governments, of UNHCR South Africa and Mozambique, of SACBC, MSF, IOM and Care International had taken part. The South African Government and UNHCR had also agreed on a figure of 250,000 to 300,000 Mozambican refugees in South Africa - accepting, in other words, the Church assessments - and had made the assumption for planning purposes that, of these, 100,000 might be expected to return to Mozambique.

(147) The UNHCR Johannesburg Office also reached an 'ad hoc' determination, following the early November protection survey, that Mozambicans who had arrived in South Africa between 1986 and 1992 would be considered as 'prima facie' of concern to UNHCR, that those who arrived between 1980-85 should be legally assessed to determine their status, while those who arrived before 1980 or after 1992 were not of UNHCR's concern. This fits in pretty well with the SACBC findings that the great majority had arrived in South Africa over a peak period during the middle and late 1980s and little, if any, assessment of earlier arrivals seems to have been required.

(148) Given the planning assumptions about the numbers to be repatriated, the operational plan was a practical and sensible document and was approved with little variation at a meeting of the Technical Committee of the Tripartite Commission in Johannesburg on 13 December 1993. This was a key meeting for the future course of the operation and was attended by no less than four representatives of the South African Government, six from the Mozambique Government, four from UNHCR Mozambique and three from UNHCR South Africa.

(149) In its essentials, the approved plan involved the establishment of up to six staging areas in South Africa where refugees wishing to repatriate would assemble for an overnight stay prior to being transported by IOM South Africa by bus, truck and possibly train to transit facilities in Mozambique. From there IOM Mozambique would take them on to their final destinations if these were beyond walking distance (20km). MSF(France) would be responsible as UNHCR's implementing partner for the construction of the staging areas in South Africa to a capacity of up to 500 persons, for pre-departure health checks, for water and sanitation and for the provision of an evening meal the day before departure. Finally, MSF(France) would conduct a mine awareness campaign. A comprehensive information campaign, involving a magazine, posters, videos and radio would be mounted, followed by mass registration by specially trained teams. The Masungulo Agency (as the implementing arm of SACBC was now called) would conduct the information campaign and registration. It would also undertake the distribution of a month's food supply49 prior to departure along with non-food items such as blankets, buckets and plastic sheeting.

(150) On the Mozambican side, transit facilities would be established using NGO partners at selected border crossing points and at district centres in areas of major expected return. These facilities would have a capacity of up to 500 persons but be capable of extension; normally, repatriants in transit would be there for only an overnight stay. Five border crossing points were agreed, to be used as and when required (one subject to de-mining). UNHCR would establish sub and Field Offices to oversee the process on both sides of the border.

(151) For immigration purposes, the VRF would be the only required document, and customs procedures - save for motor vehicles - would be waived. Significantly for the future of the programme, the South African authorities reported their acceptance that 'returning refugees would be allowed to take their personal belongings including building materials with them'. They could also take crops and small domestic animals but larger livestock would have to be herded across by the refugees themselves. No baggage allowance limits were expressed.

(152) The registration and movement of returnees would take place in two phases. Under the first phase, scheduled to commence on or about 10 January 1994, UNHCR would assume responsibility for what the Technical Committee now termed "the assisted spontaneous repatriation movement currently facilitated by the South African Defence Forces". There seems no question but that the SADF had been assisting spontaneous returnees in a humane and efficient manner and had certainly helped to save lives by rescuing them from a hazardous passage on foot across the Kruger National Park.

(153) Yet it was clearly an embarrassment to UNHCR that, with its status in the country regularised since September, its normal functions should still be carried out by the armed forces of the country of asylum. Accordingly, the Masungulo Agency as implementing partner would register the spontaneous assisted as from 10 January and issue them with VRFs, and IOM would undertake their transportation. By mid-March 1994 the repatriation of 1,925 persons had been facilitated in this way, with a regular UNHCR field presence to oversee the movements. The second phase of organised voluntary repatriation would begin at the end of April 199450.

(154) In preparation for this phase, UNHCR offices were established at Giyani, Phalaborwa, Nelspruit and Winterveldt (North Pretoria) and international staff appointed between January and March 1994. Emergency team support from Headquarters was invaluable in setting up these offices. Five staging areas were ready to open in April. The information campaign was under way. In Mozambique, eight (later eleven) transit facilities were identified of which four were operational to a capacity of 500 by mid-March and the remainder expected to be operational by the end of April. The upgrading of two of these facilities to a capacity of 1,000 was being undertaken. Three UNHCR Field Offices at Chokwe, Ressano Garcia and Chicualacuala were open or about to open.

(155) The record of activity following the 13 December 1993 meeting was, on both sides of the border, remarkable by any standards. At the second Technical Committee meeting at Nelspruit on 16 March 1994, it could finally be reported that most of the recommendations of the first meeting had been implemented. A target date of 4 April for the beginning of organised repatriation was now set. Despite the difficulties inherent in, for example, reaching so dispersed a population for information and registration or of knowing in advance how many were likely to arrive at a staging area for repatriation, the operation was ready to begin and - apart from some initial difficulty over repatriation packages on the South African side - the necessary infrastructure was in place.

THE OPERATION

(156) From the very beginning, the number of returnees was well below a projected rate of 10,000 per month or 2,500 per week up to the end of 1994, with figures of a mere 400 per week being regularly recorded. By 24 June 1994, only 6,208 persons had been repatriated against a projection of 30,000 and UNHCR South Africa was already anticipating a reduction in total numbers from 100,000 to 40,000 or even lower. Sub-agreements were being reviewed, although expenditures on most capital infrastructure could not now be reduced. The idea of some repatriation taking place by train was abandoned.

(157) By the time of the Third Technical Committee meeting at Maputo on 19 August, the number of organised repatriants stood at 11,285 out of 29,000 who had registered to return. The Committee responded by recommending (1) a reduction in the caseload from an original estimate of 250,000 to 120,00051, (2) a cut-off date for registration of 31 December 1994, with a proviso that families, as opposed to individuals, who reported to staging areas for repatriation after the date would nevertheless be assisted and (3) the closure of the operation at the end of March 1995 (with a possible review and extension to the end of May).

(158) An undated Declaration by the three parties to the Tripartite Agreement approved these recommendations. Despite some increase in the numbers repatriating towards the end of the year, due probably to the successful outcome of the Mozambique elections in October and the impending end of registration, there was no dramatic improvement and, by the end of 1994, a total of only 21,309 had been assisted to repatriate, at an average rate of 418 per week. By December 1994, the number of transit centres in Mozambique had been reduced to three.

(159) By late 1994, moreover, the so-called 'revolving door' phenomenon had become one of serious concern. The UNHCR Mozambique Field Office South was reporting figures of 65% single men per convoy. Two November convoys are on record as comprising 41 single men out of a 44 total and 103 out of 125 respectively. From 1 December, therefore, by agreement between both UNHCR offices, the repatriation of single men, nearly all of whom were simply taking advantage of free holiday travel and often had no baggage, was generally stopped. How many returnees in all went back to South Africa at some time after their repatriation to Mozambique is still impossible to say.

(160) By September 1994, virtually every Mozambican in South Africa knew that the programme would end by 31 March 1995 (in fact the last movement took place on 6 April). The result was a significant acceleration in the last three months with 10,701 returning, of whom nearly eight thousand moved in March and early April. The grand total for the two years 1994-95 was 32,010. Figures for spontaneous repatriates at least as great as those returning with assistance from UNHCR have been compiled by NAR (the Mozambican Governments refugee department, the Nucleo de Apoio cos Refugiados) but are impossible to verify.

(161) How many Mozambican refugees are left in South Africa? Again, the figure is quite impossible to verify but informed guesses put it at less than 200,000, perhaps a great deal less. Once South Africa passes its refugee legislation, the invocation of a cessation clause may conclude the matter once and for all although it will certainly not put an end to the problem of migration, legal or illegal, to South Africa from Mozambique.

TRANSPORT

(162) There is resentment by some field staff that IOM was chosen as an operational partner for the transportation of returnees by Headquarters rather than in the field, a resentment which is exacerbated by the understanding that IOM has a 'blanket' agreement negotiated at Headquarters which exempts it from the reporting requirements which are among the mandatory clauses in UNHCR's standard form of agency agreement52. This is certainly very strange and, whatever the reasons which prompted the selection of IOM for this operation, the choice of implementing partner is surely best left to the field office to determine.

(163) There is also among some a feeling that, in a country where road transport is freely available, UNHCR could more economically have contracted directly rather than through an implementing agency. Widely different figures are quoted on this point from US$180 per head to IOM's figure of US$21 per head as against US$7.5 budgeted which it attributes to 'the excessive amounts of luggage carried home by the returnees'. The same IOM report53 quotes instances where an eight ton truck would be filled by only one family's household goods. In effect, the refugees took with them all that they had in their dwellings including timber and metal sheets for roofing, doors/window frames, cupboards, bicycles, etc, etc.

(164) There is reference here and there in the documentation to 'reasonable amounts of luggage' and, in the record of a Cross Border Coordination meeting at White River on 19-20 December 1994, to the original assumption of 250 kgs per family proving unrealistic. Indeed, there is general agreement that the amount of luggage transported and the cost reached unprecedentedly high levels. Anecdotal evidence speaks of 3 buses followed by 33 trucks carrying luggage in one convoy and there is a confirmed report of 34 trucks transporting 24 returnees and their effects in one of the last convoys on 31 March 1995 (the two convoys might be the same). IOM reported the average for 1994 as one truck of belongings per 30 refugees.

(165) The lesson to be learnt is, quite clearly, that in future repatriation operations reasonable luggage limits (and what these are can only be determined in the local context) should be set and, as nearly as possible, adhered to. At the same time, it must be said that (1) the South African situation was probably unique, with refugees elsewhere (save in Swaziland) having far fewer belongings to transport, (2) a less liberal approach to the luggage question would probably have resulted in an even smaller number of refugees electing to return home and, most importantly, (3) in a society as impoverished as Mozambique, the accumulated possessions of the returnees must have been of general benefit and an aid to the process of reintegration.

(166) On the wider issue of contracting directly, it is easy, one fears, to underestimate the expertise required to supervise and administer a transport operation, even using hired vehicles from presumably experienced commercial operators. Better leave it to those whose proper task it is, even (which may be uncertain) at extra expense and concentrate on our own proper task which is protecting and assisting refugees.

(167) The transport operation was not a simple one. Convoy routes of up to 515 kilometres were involved. Staging areas were far apart and local pick-up arrangements as well as cross-border travel was involved. One convoy route had to use boats to cross the Limpopo river.

(168) The basic problem, for which IOM was not responsible, was the lack of a proper link between registration and repatriation. By the end of 1994, 53,812 refugees had registered for repatriation but only 21,309 had actually returned. In a real sense, this was less an organised repatriation than an assisted spontaneous one with planning almost impossible since the refugees only reported the evening before departure and it was impossible to know till then how many would be going. This led to either a shortage or an oversupply of vehicles and an inability to advise the Mozambican side in advance how many would be coming. The only answer is, of course, to register at the point of departure and not before.

KEY QUESTIONS

(169) Were returnee estimates justifiable and realistic? In the circumstances, yes. Neither the South African Government nor most certainly, UNHCR had any first-hand reliable information in advance. The vast majority of the refugees had settled in the former so-called 'homelands' of Gazankulu and Kangwane in the Northern and Eastern Transvaal or in the former Bophuthatswana and little was known about them. UNHCR had no right of access to them until the Basic Agreement was signed in September 1993. There was little choice but to accept the South African Council of Churches' estimate of 250,000, while some thought that up to 300,000 might be involved. The SACBC Survey - which UNHCR helped to finance - concluded that over eighty percent of the refugees wanted to return to Mozambique and from what was known at that time (late 1993) about repatriation from other countries of asylum, this must have seemed a reasonable figure.

(170) Indeed, the estimate that 120,000 would need to be assisted was a modest rather than an extravagant assessment.

(171) From other countries of asylum, the proportion of refugees returning voluntarily to Mozambique ranged from the majority to virtually the entire caseload. South Africa was the exception that proved the rule, a conclusion based on hindsight, not on the information available at the time. For example, in a paper presented by UNHCR Mozambique's Field Office for the Southern Region to the November 1993 Inter-Agency Needs Assessment Mission, a figure was given of 35,000 spontaneous returnees who had already arrived back in the Gaza, Maputo and Inhambane Provinces with the comment that this 'indicated the readiness of Mozambicans to return to their country'. Elsewhere in the region that readiness was only too apparent.

(172) The only lesson to be learnt from this experience is, wherever possible (and in this instance it was not possible) - do your own surveys, investigations and forward planning well in advance of the actual commencement of operations. It can be argued that, if the starting point for repatriation from South Africa had been delayed until, say, the middle of 1994 a much more accurate estimate of potential returnee numbers could have been obtained. But this would have been achieved at the cost of impaired relations with both governments, a rising tide of xenophobia and a loss of credibility as the SADF went on doing UNHCR's job for it in the Kruger Park. There was no real alternative except to take the best available estimate and begin the operation at the earliest practicable moment. In the outcome, little if anything was lost, except a certain amount of wasted effort.

Why did so few refugees return from South Africa to Mozambique?

(173) If the initial estimate of 250,000 to 300,000 refugees was correct, then no more than 15% to 20% availed themselves of the opportunity for assisted organised repatriation. Nor can one be sure how many of those returning to Mozambique were genuine refugees and how many economic migrants taking advantage of a generous return package or the opportunity for an easy turn through the 'revolving door'. Many of the single young men who became such a prominent feature of convoys in the last months of 1994 must surely have returned to South Africa after a brief visit to Mozambique. Others who registered for repatriation did so only as a personal insurance against deportation and never returned to Mozambique at all. Throughout the operation more Mozambicans were probably arriving in South Africa - legally or illegally - than were returning to Mozambique. The Regional Office Johannesburg states that the South African authorities deported some 80,000 Mozambicans in 1994 while UNHCR repatriated 22,000 - a ratio of deportations to repatriations of almost four to one!

(174) A difficult background then, to organised repatriation. The main reason for the paucity in numbers must have been, however, that life was so much better in South Africa than in Mozambique and that few believed that conditions in Mozambique were actually improving. In this belief, they may have been sustained by information (or misinformation) from South African employers who wanted to keep them as cheap labour in farms or mines.

(175) Other reasons, a fear of returning before the elections, difficulty in reaching staging areas and so on have been adduced. None is convincing - there was no great upsurge after the elections and the provision of trucks from pick-up points to staging areas did not result in an improvement. There is a sense, however, from those most closely involved on the part of the Government and UNHCR, that the SACBC Masungulo Agency was not the ideal operational partner, did not conduct the information campaign as comprehensively and efficiently as had been hoped and, most importantly, lacked real enthusiasm for the cause of repatriation despite its Survey finding that over eighty percent of households questioned wanted to return to Mozambique. With the benefit of hindsight they feel that it was the wrong agency to choose and too closely linked to church influence from Mozambique.

(176) But there was very little, if any, choice available at the time. The South African Red Cross declined to get involved because of over-commitment elsewhere and NGOs generally in South Africa were not yet enthusiastic about working in close relationship with governmental authorities, having been historically so much in opposition to the government in the past. So Masungulo was really the only option.

SELF EVALUATION

(177) A joint UNHCR South Africa/UNHCR Mozambique evaluation meeting to review the repatriation operation was held in South Africa on 20 and 21 April 1995, another commendable example of the various evaluation initiatives which have been taken in the region54. The writer of this report was privileged to be able to attend. There were 12 participants in all, including representatives from all the Sub and Field offices on both sides of the border and the Deputy Representatives from the Regional Office in South Africa and the Branch Office in Mozambique. The agenda had been jointly prepared.

(178) The recommendations or lessons learnt from the meeting are contained in a five-page report, the salient points from which can be summarised as follows:

(a) UNHCR's field presence should have been established earlier or at least a more detailed analysis of the caseload undertaken well in advance. Operational planning was adversely affected by the lack of accurate information about the refugees.

(b) Registration and departure should have been directly linked. The same organization should not have been responsible for both the information campaign and the subsequent registration.

(c) Operational autonomy by field offices on both sides of the border, direct communication by PACTOR/SITOR, VHF radio and other means along with cross-border travel authorisation and simplified immigration were key factors in success.

(d) The Technical Committee formula, embracing both Governments and UNHCR on both sides, was a valuable instrument for defining policy and directing implementation.

(e) For organised repatriation involving more than one country of asylum, standardised repatriation packages should be developed, especially for food items.

(f) Implementing partners (eg. IOM) should be selected not at Headquarters but in the field.

(g) Luggage allowances should be related to budgetary provision and be logistically manageable.

(h) The operation was justified. Both Governments requested UNHCR assistance and UNHCR has a mandatory responsibility to assist refugees who wish to return home. Repatriation took place in a voluntary, safe and dignified manner.

(179) Many other points of detail were covered at the meeting and recommendations made. The key thrust - as in other countries of asylum - is that the operation would have been more effective had preparations begun earlier and an earlier UNHCR presence been established in the field. A separate 'lessons learnt' paper by the UNHCR Field Officer at Giyani55 dated 23.3.95 reiterates the point and suggests that the start of the operation might usefully have been delayed to allow adequate preparations to be made. He also poses the question whether the refugees should have been given such complete freedom when to repatriate (within the overall timetable for the operation) rather than instructions to each settlement or group of settlements that a specific period of one month had been set aside for their repatriation preceded by a concentrated registration effort. The pace of repatriation would thus have been controlled.

(180) There seems to be general criticism of the allegedly rather dilatory manner in which registration was carried out and some misgivings by protection staff in Johannesburg that registration was entrusted to an NGO rather than retained as a core function of UNHCR. Others feel that, since the voluntary nature of the repatriation was not an issue, registration by an NGO was acceptable here as, indeed, elsewhere. One is inclined to agree.

CONCLUSION

(181) Was the operation a success? On its own terms, unequivocally yes, with the refugees returned to Mozambique in a safe, dignified and, in the circumstances, generous fashion. UNHCR's obligation to both governments was satisfactorily fulfilled and, in the case of South Africa, vital steps were taken towards the establishment and maintenance of good relations with the authorities which may well prove invaluable in the future. Certainly, the principal government official involved in the repatriation programme has nothing but praise for the relationship with UNHCR.

(182) Related to the operation has also been the training of government immigration officials and others, the preparation of draft refugee legislation and generally the creation of a secure base from which the UNHCR Regional Office can maintain its presence in the future.

(183) On the debit side, much was inherently unsatisfactory about the operation from the start. The caseload was an unknown quantity and only the roughest measure of eligibility could be applied to determine who was a genuine refugee and who was not. There was no guarantee that the more vigorous of those repatriating might not return to South Africa as economic migrants after leaving property and older family members in Mozambique at UNHCR's expense. There was no clear notion at the end of the exercise of how many refugees were left in South Africa, pending either spontaneous repatriation, a cessation clause or deportation. Some funds, although this should not be exaggerated since staging and transit facilities were always modest, were undoubtedly wasted in providing for many more returnees than actually appeared. The impact on the problem of Mozambicans in South Africa was negligible, although the return of so much property - and some relatively skilled people - to Mozambique must have had some beneficial effect.

(184) Not, then, the kind of operation that UNHCR should willingly embark upon unless obliged to. In this case, it was obliged to and fulfilled the obligation on both sides of the border efficiently and well. And with its eyes open - the May 1993 Plan of Operation is frank about the lack of extensive data and the dangers of the 'revolving door'. It also expresses concern about the possibility of forcible repatriation which, had no repatriation exercise been mounted, might well have taken place.

SWAZILAND

(185) The March 1993 Plan of Operation gave the number of UNHCR assisted Mozambican refugees in Swaziland as 24,00056, living and, to a limited extent, farming at the two refugee camps of Malindza (17,000) and Ndzevane (7,000). The refugees came primarily from the southern Mozambique Provinces of Maputo and Gaza and had arrived in successive influxes between late 1984 and the middle of 1992. A further 40,000 refugees were believed to be spontaneously settled along the Swaziland/Mozambique border and were expected to return home without assistance. Assisted repatriation would be implemented over a period of six months to a year from August/September 1993 onwards.

(186) The operation was completed well within the year. A Tripartite Agreement in the usual terms was signed on 20 August 1993 and the operation began, with IOM as logistics partner, on 14 October 1993 with an initial convoy of 519 returnees. The operation was completed on 3 June 1994, by which time a total of 16,666 had returned on special passenger trains to reception centres at Umpala or Matuba railway stations and onward transportation to their homes.

(187) By the time of the evaluation mission to Southern Africa, the operation had been over for almost a year, the Branch Office in Swaziland had been closed and residual refugee matters had come under the authority of the newly established Regional Office in Johannesburg. There was no reason to justify an evaluation visit to Swaziland and little about the operation that calls for extended comment. The number of those repatriating was relatively small, the areas to which they returned, such as Boane District, were mostly close to the Swazi border and full advantage was taken of the logistic opportunity to transport the refugees by rail.

(188) The only aspect of the operation that calls for particular comment is that of the baggage allowance and so-called 'repatriation package' for returnees. The baggage entitlement was set at 50kgs per person or 250 kgs per family (the same as subsequently in South Africa) but this limit, as in South Africa, seems to have been honoured more in the breach than in the observance - although, in fairness, the refugees were told that flexibility would be exercised to facilitate the transportation of most of their property. Each family was also permitted to take six wooden poles and iron sheeting from their homes to use for reconstruction in Mozambique.

(189) The 'repatriation package' included a 200 litre drum, household utensils, building and agricultural tools, building materials and seeds. Two weeks' rations were issued before departure. The refugees were told in the course of an admirably concise but comprehensive information leaflet that they would be expected to construct homes in Mozambique with their repatriation kit.

(190) There has been some criticism of these allowances as unduly lavish and they were undoubtedly liberal. The repatriation package certainly contained more than in any other country of asylum. But the emphasis on reconstruction and house-building was admirable and may well have contributed greatly to successful reintegration. The availability of rail transport, which was not used elsewhere except Malawi, and there only to a very limited extent, made generosity feasible57 as did the relatively short distances from reception centres to the final destination of the great majority of refugees.

(191) It does seem, however, that very little effort was made to coordinate repatriation packages and baggage allowances between different countries of asylum, with each Branch Office apparently using its own judgement over how much to provide. Complete uniformity would, of course, have been impossible - the need for an initial food supply for refugees returning to Cabo Delgado from Tanzania was obviously much greater than for those coming from Swaziland to Maputo while, conversely, logistic difficulties dictated much stricter limits on the baggage allowance for returnees to northern Mozambique. Nevertheless, more could probably have been done to achieve, if not uniformity, greater consistency and this should be borne in mind for future operations.

STATISTICS

(192) The Swaziland statistics are more than usually confused. UNHCR sources give the number of assisted refugees as variously 24,000, 18,000 or even less ("some 16,400 were repatriated over an 8-month period. Approximately, 200 refugees remain in the camps". Repatriation and Reintegration Interim Progress Report June 1994). The February 1994 Fund Raising Appeal gives the Swaziland caseload as 18,000 at 1 January 1994, all 18,000 planned to be repatriated in 1994. The January 1994 Repatriation and Reintegration Progress Report maintains the earlier figure of 24,000 but says that 5,970 had been repatriated by the end of 1993. If 5,970 is deducted from 24,000 it leaves about 18,000 to be repatriated in 1994 but, since only 10,696 were (presumably) repatriated in 1994 to make up a grand total of 16,666 what happened to the rest? To add to the confusion, the UNHCR Special Report 'Mozambique: Back on Track' of February 1995 says that 11,600 were transported between January and May 1994 which, added to 5,970 in 1993 would give a total for the operation of 17,750.

(193) The estimated number of spontaneously settled refugees in Swaziland is 40,000 in May 1993 but by January 1994 has gone down to 30,000. An October 1993 Information Bulletin gives a total for all Mozambican refugees in Swaziland of 48,100 which is consistent with none of the other figures current at that time.

(194) The Tripartite Agreement is reported as having been signed on either 19 or 20 August 1993. There were either 25 or 29 convoys repatriating either 16,400 or 16,666, according to UNHCR or 16,484 according to IOM. The operation was completed on either 3 or 6 June 1994.

(195) This may appear merely a petty irritant but UNHCR programme statistics have long been plagued with anomalies and contradictions; there are other examples from the Mozambique operation which have been left unquoted. Credibility can be affected by figures which do not appear to tally and more effort should be made to achieve consistency. Should a Voluntary Repatriation Unit at Headquarters be established, this could be one of its many objectives. One fundamental problem is the multiplicity of organizational units producing documents with statistical information - the Bureau, the Fund Raising Service, the Public Information Section, Regional and Branch Offices as well as IOM and other implementing partners.

(196) Meanwhile, on paper, at least, the question remains. If there were 24,000 refugees in two camps when the operation started, if 16,666 (or some such figure) were repatriated and 200 were left, what became of the other 7,234?

TANZANIA

HISTORY AND BACKGROUND

(197) The first influx of Mozambican refugees into Tanzania took place in the 1960s and early 1970s during the war of independence from Portuguese colonial rule. Some 50,000 of these refugees were accommodated in agricultural settlements, from which 20,000 were repatriated to Mozambique in a process which was completed in late 1976. Others remained in Tanzania and it is said that there are 300,000 or more Mozambicans in Tanzania to this day, many of the same ethnic origin - Wayao and Makonde - as the Tanzanians among whom they live.

(198) The second influx, and the one with which this evaluation is concerned, began in 1986 with the spread northwards of the civil war between the FRELIMO Government and the RENAMO opposition in Mozambique. In 1989, a settlement was established at Likuyu, 110 kilometres northeast of Songea in the Rovuma region for those Mozambican refugees who were not spontaneously settled but had hitherto been accommodated in reception centres within a radius of 50 kilometres from the Mozambican border. A UNHCR Sub-Office was opened at Songea in 1988 but was closed in 1992, allegedly as a 'trade-off' for the establishment of another Sub-Office at Kigoma. By 1993, there were some 18,400 Mozambican refugees at Likuyu, the last of whom, numbering about 3,000, had arrived as late as mid-1992. The refugees, who were predominantly of peasant subsistence-farming origin, were divided between nine villages with between 1 and 2½ hectares of land per family and had mostly become self-sufficient within two or three years of their arrival.

(199) From the outset, UNHCR had two main NGO operational partners at Likuyu, the Tanganyika Christian Refugee Service (TCRS) and the Tanzania Mozambique Friendship Association (TAMOFA). TCRS, which has a long history of association with UNHCR in Tanzania and vast experience of rural settlements for refugees, was responsible for the development of the infrastructure , agricultural inputs, distribution of both food and non-food assistance and for basic services. TAMOFA, founded in 1988, implemented project assistance in education, health, income-generation and community development. Both agencies operated under tripartite agreements with the Tanzanian Ministry of Home Affairs and UNHCR.

(200) Commonly accepted estimates have given the number of spontaneously settled Mozambican refugees in the southern Tanzanian province of Mtwara as about 57,000 by 1992-93. UNHCR Mozambique Repatriation Progress Report of 27 April 1995 cites a figure of 50,000 and says that some58 14,882 had repatriated by the end of March 1995 with no assistance from UNHCR and estimates a further 20,000 possible returnees in 1995. Other observers consider the total an exaggeration; the TCRS Director in Tanzania believes the number to be not above 20,000 and points out that local people are reluctant to identify the spontaneously settled refugees who live with them and share the same ethnic roots.

(201) As indicated earlier, the total number of Mozambicans living in Tanzania is probably many times greater than the number of refugees. In any event, there is general agreement that the Tanzanian authorities, in keeping with their traditional hospitality towards refugees, welcomed the Mozambicans and were quick to provide them with land on which to settle. The settlement facilities have benefited local people and the presence in their midst of refugees has not engendered conflict or resentment. The Government has not exerted pressure for their return to Mozambique and has extended the option of naturalisation to those who wish to remain permanently in Tanzania. Likuyu settlement has been a model of its kind and the refugee community has prospered through the production and marketing of tobacco and other crops.

(202) From this background, several relevant points emerge. The planning and execution of an organised repatriation programme benefited greatly from (I) a relatively small number of refugees eligible for repatriation from an orderly and well administered land settlement, (ii) the established presence on the settlement of no more than two NGO implementing partners who were between them responsible for the whole range of services to the refugee community and were available to lend their knowledge and expertise to the repatriation at both central and local level.

(203) On the other hand, the logistic problems of organised repatriation to Mozambique from Tanzania were likely to be greater than from any other country of asylum in the region. The Likuyu settlement was 110 kilometres and three hours drive from Songea, the nearest town of any size. The settlement itself was spread out over 27,000 hectares and at distances of several hundred kilometres along bad roads from potential crossing points into Mozambique; the shortest route to Lake Malawi at Mbamba Bay was 264 km and a ten-hour drive. The two Mozambican provinces to which the great majority of the refugees would have to be repatriated - Niassa and Cabo Delgado59 - were (and remain) the most remote in Mozambique. The problems of communication and infrastructure for the reception and reintegration of returnees in their areas of origin were daunting and remained so throughout the operation.

PLANNING AND PREPARATION

(204) An initial Plan of Operation for voluntary repatriation was prepared by the Branch Office in March 1992 and was followed by a project description and budget submitted to Headquarters on 22 December 1992 for a repatriation project to run from 1.3.93 to 28.2.94. This proposed the appointment of two consultants, one (local) to develop a registration system and one (international) to be the field focal point during the initial months of the operation. A few days later, on 30 December 1992, a formal request for the creation of posts for the re-opening of the Songea Sub-Office as a Field Office was sent to the responsible Desk IV at Headquarters, asking for one international staff member as Head of Field Office, one National Officer and supporting GL staff. Job descriptions were provided and the posts were requested from 1.2.93 for the programme to run until 28.2.94. It was envisaged from the inception of the project that UNHCR's operational partners in the repatriation would be TCRS for the logistics element - transport, transit centres and roads - and TAMOFA for health services, information, registration and community support.

(205) There followed a lull. The creation of posts, in the opinion of some field staff, 'fell between the cracks' separating Desk IV in the Africa Bureau from Desk V, which was responsible for all the other countries of asylum in the region. Be this as it may, it was clearly also realised that a considerable amount of preliminary work needed to be done before the programme could be launched, including a great deal of repair and rehabilitation in Mozambique itself.

(206) This need was reflected in the first information campaign mounted by the Branch Office in May and June 1993 and the production at about that time (the document is undated) of a first information pamphlet circulated to the refugees in English and Swahili.60 The refugees were told that repatriation could only start in mid-1994, after the 1994 harvest at Likuyu, and that a number of conditions would have to be met in Mozambique before any movement could take place. These conditions included safety through peace and stability, the restoration of basic services such as health, education and local administration and the availability of tools and seeds. The refugees were told that they would travel home with transport organised by UNHCR and would be assisted during the journey.

(207) The terms of this information pamphlet were criticized by the TCRS Director in Dar-Es-Salaam at a meeting with the evaluation mission on 13 March 1995 at which he characterised as misleading the statements that basic service would be restored in advance of repatriation and that UNHCR could as he put it, 'take the refugees all the way home'. The assurance over basic services could certainly be considered rash, particularly in relation to remote areas where, even without the effects of civil war, they had hardly existed in the first place. Nor, given the appalling state of road communications in northern Mozambique, could all the returnees be escorted to their homes.

(208) The TCRS Director did not, however, seem to be aware that this information pamphlet was superseded and replaced through a second information campaign mounted a year later in May and June 1994, with an updated pamphlet and radio broadcasts directed at the refugees. Both pamphlet and broadcasts considerably toned down the assurance over services in the following terms:

schools, clinics, water pumps and other infrastructure will be rehabilitated by the Government of Mozambique, the UN and NGOs in accordance with the National Reconstruction Plan. However, emergency infrastructure and basic services will be provided in returnee resettlement areas.

(209) The document went on to say that refugees would be assisted by UNHCR, TAMOFA and TCRS to reach crossing points in Tanzania, from where they would travel to their final destinations in Mozambique. UNHCR would ensure safe travel to the final destination and basic material assistance - agricultural seeds and tools and food until the next harvest - would be provided. The tools and seeds would be distributed at Likuyu prior to departure.

(210) This question of successive information campaigns and whether the first of these in 1993 could have misled the refugees has been dealt with in some detail (and out of strict chronological order) because it constitutes a rare example of direct criticism by an NGO operational partner of UNHCR's role in the repatriation to Mozambique. The first pamphlet was undoubtedly optimistic and promised more inside Mozambique than either UNHCR or the Mozambican authorities were in a position to provide. Nevertheless, the second version, a year later, close to the commencement of the actual repatriation and supplemented by radio broadcasts in identical terms to the printed document, was a fair and helpful statement of what could be done and must have replaced any earlier impression in the refugees' mind. It was, moreover, approved by the Tripartite Commission (see below) in May 1994 and supplemented by information visits to Likuyu.

(211) The first information campaign was followed in July 1993 by a UNHCR-conducted socio-economic survey aimed at gathering basic data to ascertain which of the refugees wanted to repatriate, where they came from in Mozambique, their family composition and skills profiles and other information of this kind. The results showed that 17,675 or about 90% of the Likuyu population wanted to be repatriated, 17,003 to Niassa and 672 to Cabo Delgado. A handful of refugees from other provinces, 49 in all, were also listed. It is interesting to note that there was apparently a trickle of spontaneous repatriation, numbering 300 to 500 persons, from Likuyu to Mozambique in 1993 but that some of these returned to Likuyu saying that 'conditions were bad and nothing was available in Mozambique'.61

(212) The Tripartite Agreement between the two governments and UNHCR for voluntary repatriation was signed on 11 November 1993 at Dar-Es-Salaam and contained the usual - and extremely valuable - provisions for the use of Voluntary Repatriation Forms (VRFs) or other simplified procedures, two-way cross-border travel authorisations for the staff of UNHCR and implementing agencies, multiple visas for drivers and appropriate mechanisms to facilitate the conversion of local currency. A small point perhaps but no names are printed to identify the signatories and the UNHCR signature is illegible. The signatories to any formal agreement of this kind should be clearly identified, preferably by official title as well as name.

(213) TCRS had already conducted a physical assessment of possible routes and crossing points in August 1993 and, now that an official declaration had been made in the form of the Tripartite Agreement that repatriation would take place, formal logistic and organisational planning could begin. A Needs Assessment Mission, composed of representatives of UNHCR, IOM and TCRS visited Cabo Delgado and Niassa Provinces between 22 and 27 November 1993 and recommended that five routes for repatriation should be used - via Lake Malawi for returnees to Lago, Lichinga and Muembe Districts in Niassa Province, from Mtwara by sea to Mocimboa da Praia or Pemba for returnees to Cabo Delgado and three overland routes via Matchedje, Millepa and Gomba respectively, subject to further technical assessment. TCRS would organise movements on the Tanzanian side, while IOM would be responsible for the lake voyage and all inland transportation within Mozambique.

(214) The Tripartite Commission held its first meeting in Maputo on 6 December 1993. Generally, it endorsed the Needs Assessment Mission's conclusions but noted that major road works would be necessary on the Tanzanian side to link the three border crossing points with the Tanzanian road network, that there were no bridges across the Rovuma river and that the roads within Mozambique to Mavago, Mecula and Metangula would need to be upgraded (probably the understatement of the day). The Commission also recommended that UNHCR and the Mozambican government agency NAR (Nucleo da Apoio nos Refugiados) should establish a presence in Niassa and Cabo Delgado Provinces without further delay.

(215) The question of routes was further explored during a mission to Tanzania between 18 and 22 January 1994 by the Deputy Director of IOM Mozambique, including a visit to the Likuyu settlement with the UNHCR Deputy Representative for Tanzania. In summary, they concluded as follows (taking their figures from the June 1993 socio-economic survey):

PROPOSAL 1. Lake Malawi Route

MBAMBA BAY (Tanzania) to METANGULA (Mozambique)

(216) A total of 4,565 persons to be transported, which would take 30 4-day trips by available landing craft on 121 days. IOM to look into the possibility of using Malawi lake ferries which could take more passengers on fewer trips.

FACT A total of either 8978, 9298, 9301 or 9008 (differing figures variously quoted by TCRS or UNHCR sources) actually returned to Mozambique by this route on 31 convoys by MV Songea and MV Mtendere between 14 July and 30 September 1994 - the so called "Operation Sunset" which carried by far the greatest number of repatriating refugees. Mbamba Bay is 264 km from Likuyu, the road journey lasting 10 hours.

PROPOSAL 2. Sea Route

MTWARA (Malawi) to MOCIMBOA DA PRAIA (Mozambique)

(217) The mission concluded that the need for this route be looked at again by IOM and TCRS, as the majority of refugees from Cabo Delgado came from areas close to a possible crossing point near Newala in Tanzania and could go by land and river crossing - TCRS and IOM to examine the feasibility of this alternative.

FACT In all, 323 refugees (as against an anticipated 672) refugees were repatriated by sea from Mtwara to Mocimboa da Praia on 10 to 12 September 1994. The alternative land route was never used. Mtwara is 690 km from Likuyu by road.

PROPOSAL 3. Land Route

MASUGURU (Tanzania) to GOMBA (Mozambique)

(218) IOM was asked to look into the possibility of using two suggested alternatives to this route, which might be more direct and secure for the majority of the 3,548 persons expected to return to Mecula District.

FACT Two thousand and eighty refugees were repatriated by this route on 10 convoys between 29 September and 21 October 1994. The possible alternative routes were never pursued and, indeed, this was the only land route used on any scale.

PROPOSAL 4. Land Route

MITUMONI (Tanzania) to MATCHEDJE - MACALOGE (Mozambique)

(219) The mission proposed that, since all but 63 of the 2433 persons expected to use this route came from the south of Sanga District i.e. the Macaloge area, it might be more sensible to take them by the lake route and then home over relatively good roads.

FACT Except for "a couple of persons" (TCRS), the route was never used, although the Tanzanian regional authorities made efforts to clear a road to the river and the possibility of using it seems not to have been finally excluded until early September 1994.

PROPOSAL 5. Land Route

MAGAZIN (Tanzania) to MILEPA (Mozambique)

(220) The mission considered, based again on the June 1993 survey, that this crossing would have to cater for the largest population movement of all, numbering 6,372 persons. It noted, however, that (1) transit centres would be needed on both sides of the border, (2) substantial road and bridge works would be needed in Tanzania, including the construction of 20 km of road between Magazin and the Ruvuma river and (3) that a report on the feasibility of using the road inside Mozambique between Milepa on the border and Mavago some 75 kms inland was most important. A glance at the map shows at once that the use of this road was essential to the viability of the route.

FACT Apart from a crossing by 21 persons on 21 November 1994, this route was never used. A great deal of preparatory work by TCRS on the Tanzanian side was wasted. As will be seen when we consider the implementation of the project, the problem was on the Mozambican side with work on the access road from Milepa to Mavago incomplete within the time-scale set for the operation.

(221) The time-scale was finally and firmly set by the Tripartite Commission at its second meeting (described as a Technical Committee meeting) on 21 March 1994 at Dar-Es-Salaam. Movement was to begin in June/July "to allow ample time to prepare fields for the rains in August/September and to be in Mozambique for the elections in October". Clearly, the election date was already a significant determinant in setting time limits for the operation.

(222) Planning this far had proceeded carefully and well. The March meeting of the Commission approved a Plan of Action covering movement, assistance packages and longer-term food assistance, registration proposals were set out, arrangements agreed for a Mine Awareness Campaign to be conducted by Norwegian Peoples' Aid and for mass information by TAMOFA. Currency exchange arrangements up to a limit of $5000 per head were accepted and simple immigration and customs procedures considered. UNHCR Tanzania made it clear that it was involved only in organised and assisted repatriation from Likuyu and not in spontaneous repatriation from elsewhere, although those who repatriated spontaneously would be eligible for relief in Mozambique.

(223) On the question of routes, the Commission endorsed all the five routes - lake, sea and land - which the January UNHCR/IOM Mission to Likuyu had elaborated and noted an assessment presented by IOM that the roads MATCHEDJE- MACALOGE and MILEPA-MAVAGO were suitable for use but would require rehabilitation and repair after the rainy season (which would end shortly thereafter). The Commission considered that the routes used should contribute to local development but that major road rehabilitation - given the time-frame - could not be undertaken.

(224) The Commission noted that a UNHCR Sub-Office in Lichinga, Niassa Province had been opened in February 1994. A Field Office in Pemba, Cabo Delgado Province was also opened in March.

(225) However, there was still no Field Office or UNHCR presence at Songea, although a reminder repeating the job descriptions for the Head of the Field Office and the Repatriation Officer had been sent to Headquarters (Desk IV of the Africa Bureau) on 15 December 1993. The Field Office posts were, in fact, created in March 1994 and only at that point could action to fill them be commenced62. One unfortunate consequence was that, although the Branch Office acted extremely quickly to recruit local staff for the operation, the recruitment process was more hurried than it need otherwise have been, references were not always comprehensively checked because of time constraints and, in consequence, not all those recruited turned out to be satisfactory. The time available for training was quite inadequate.

(226) The Third Tripartite Commission meeting took place at Maputo on 12-13 May 1994, with the UNHCR delegation now led by the Coordinator from UNHCR Headquarters. The Commission recommended that repatriation movements should begin in July under conditions of safety and dignity while expressing some apprehension over socio-economic conditions in areas of return and underlining the need for emergency infrastructures to be set up 'in areas of actual returnee settlement'. How this was to be done is less than clear since, in the same sentence, the Commission remarked on the lack of precise information on final destination - such information being surely a prerequisite to emergency infrastructures being created.

(227) The Commission also expressed fears that delays, particularly over the major road rehabilitation work required on the Milepa-Mavago route, might result in the repatriation of some 5,500 refugees being extended into 1995. Other recommendations by the Commission called for a visit by Mozambican authorities to Likuyu by the end of May, prior to the registration of refugees for repatriation63, an Information Campaign using the second campaign leaflet referred to earlier, and the establishment by UNHCR Mozambique of transit centres at crossing points.

IMPLEMENTATION

(228) The Songea Sub-Office was finally reopened with the arrival of consultants64 and locally-recruited staff progressively arriving May and June 1994. The Head of the Sub-Office (first contacted by UNHCR as a possible recruit in April) arrived on 28 May and by mid-June the registration and data entry systems were in place. Simultaneously with the arrival of staff, arrangements had to be made by the Branch Office for accommodation, furniture, vehicles, equipment including radios and all the other associated paraphernalia of a new operation. A fourth meeting of the Tripartite Commission (Technical Committee) was held at Likuyu on 23 and 24 June 1994 and recommended that repatriation should start on 15 July, using firstly the lake route, secondly the Milepa/Mavago land route and thirdly the Gomba/Mecula land route. A south sea route via Mocimboa da Praia for those going to Cabo Delgado Province was to be used after 15 August and all the movements should ideally, have been completed by 15 October before the rainy season - a three month timetable for the entire operation.

(229) It is a remarkable achievement that the repatriation target was virtually fully met. By 21 October 1994, a total of about 11, 400 refugees65 had been repatriated from Likuyu almost all by the three routes Likuyu-Mbamba Bay-Metangula (approximately 9,000) and Likuyu-Tunduru-Musuguru-Gomba (approximately 2,100) and Likuyu-Mtwara-Mocimboa da Praia (323). Between 15 July and 30 September, all repatriation by the lake route was completed, using firstly a Tanzanian ship, the MV Songea and later a Malawian ship, the MV Mtendere as well. The achievement is the greater for the fact that IOM, which was responsible for the lake voyage arrangements, did not establish an office in Songea until July 13, the very same day that the first returnees were collected from their villages by TCRS for the road journey to Mbamba Bay.

(230) Apart from a temporary hold-up of some ten days in mid-August due to problems over reception at Metangula (caused by the demobilisation of Mozambican soldiers) the operation went smoothly and well. This would not have been the case had the UNHCR/IOM mission to Likuyu in January 1994 not cast doubt on the feasibility of using landing craft from Metangula for the voyage and raised the possibility of using lake ferries instead. Nor could the operation have been completed by late September had the decision not been taken to supplement the use of the MV Songea (250 person capacity) with a bigger ship, the MV Mtendere (450 person capacity).

(231) Movement by sea from Mtwara to Mocimboa da Praia was accomplished in a single operation involving 323 persons between 10 and 12 September 199466. The journey from Likuyu to Mtwara of 690 km took two days with an overnight stop at a TCRS established transit centre at Tunduru, 250 kms from Likuyu. The same transit centre was used for repatriation by the only land route effectively to be used, Likuyu - Masuguru - Gomba, along which the first convoy left Likuyu on 30 September, reaching Masuguru on the afternoon of 1 October, with further convoys planned for every second day.

(232) At this point, the repatriation operation suffered its only real hitch. The Tripartite Commission (Technical Committee) had already decided at a meeting on 6 September that, because of the limited availability of trucks and the logistic constraints, it would be impracticable to conduct lake and land movements simultaneously, so that movement along this route only began once the lake and sea repatriations were complete. Nevertheless, despite agreement on the schedule by all concerned, trucks from IOM Mozambique were not ready at Gomba, the transit facilities at Gomba were poor and the 200 returnees on the first convoy had to remain in the Masuguru transit centre to be joined by a second convoy of 200 which left Likuyu according to plan on 2 October. At this point, further movements were halted as there was no capacity at Masuguru to accommodate additional returnees. The actual river crossing of people from the first and second convoys did not commence until 7 October, by which time five trucks had arrived at Gomba; movements thereafter proceeded according to plan until by 21 October 2,018 persons had been repatriated by this route and the entire organised repatriation operation was complete.

(233) In retrospect, the fact that the operation suffered nothing more than a few days' delay reflects considerable credit on those responsible on the Tanzanian side - principally TCRS and TAMOFA - for managing to accommodate and care for a transit centre population considerably larger and over a longer period than originally envisaged. This avoided the necessity for convoys to turn back to Likuyu which, as the UNHCR Sub-Office Songea report quoted later in this Chapter pointed out, could have sent 'wrong signals' to the remainder still awaiting repatriation.

(234) Also in retrospect, the fact that the operation proceeded so smoothly despite the late establishment of a UNHCR field presence in northern Mozambique, and even later at Songea, reflects the high degree of preparation and planning that had gone on over the preceding years, the excellent level of cooperation with both governments and, perhaps most of all, the active presence at Likuyu of two experienced operational partners who had been associated with the management of the settlement from its inception and who knew well in advance the part in the repatriation process that they would be called upon to play. TCRS have pointed out to the evaluation mission that their Project Coordinator at Likuyu had spent eight months in detailed planning for the repatriation before UNHCR arrived on site and should be given credit for much, if not most, of the preparatory ground work. For example, work on the transit centre for the lake route at Mbamba Bay was begun by TCRS in late April 1994 and on the Masuguru transit centre in May.

(235) Less fortunately, though through no fault of TCRS, extensive work which the agency carried out from October 1993 onwards to erect a transit centre, construct a 20 km access road to the river crossing point and rehabilitate a further 120 km of road and 14 bridges for the Magazin - Milepa route, proved in the end to be wasted expenditure, at least so far as the repatriation exercise was concerned. This route, originally envisaged as carrying the bulk of the returnees, was eventually abandoned in early September when no assurance that it would be ready in time to meet target dates (despite earlier expressions of confidence) could be given from Mozambique. At the Technical Committee meeting on 6 September, the decision was taken to use the lake route instead, TCRS closed the transit centre and handed the non-moveable infrastructure over to the local authorities.

(236) On the positive side, the late abandonment of what had originally been planned as a major crossing route demonstrated the programme's capacity for flexibility in order to meet its targets. In most other respects, planning and execution on the Tanzanian side seems to have proceeded carefully, indeed meticulously, and contingency needs were well anticipated in advance. The information campaign by leaflet and radio was reinforced by meetings attended by thousands of refugees in the presence of UNHCR, Ministry of Home Affairs, TAMOFA and TCRS representatives. Twenty-nine refugees were successfully trained in mine awareness techniques by Norwegian Peoples Aid in association with TAMOFA during May and thereafter participated as trainers in a total of 840 training sessions involving attendance by 13,240 refugees. Arrangements were made with the National Bank of Commerce for refugees to change Tanzanian currency into dollars. Although in the event only $2,337 was actually exchanged, the decision to establish what might have been a much-needed facility (and had been predicted by the Likuyu Cooperative Society as such) was sound.

(237) Finally, a potential crisis over payment for their crop to the tobacco farmers among the refugees was averted when UNHCR (Branch Office and UNHCR Headquarters) authorised a temporary advance of Tanzanian shillings 40,000,000 to provide the marketing organisation SAMCO with sufficient funds. The loan was repaid in full in September by which time the tobacco farmers had received the return on their harvest and were able to repatriate if they wished.

(238) In one important respect, however, programme objectives were not met. The 4th meeting of the Tripartite Commission (Technical Committee) at Likuyu on 23-24 June 1994 had clearly stipulated that - in view of the extreme isolation of many areas of return in northern Mozambique and associated distribution problems - a three months food ration should be provided by WFP prior to departure67 in addition to a 10-day ration to meet immediate family needs. However, due apparently to delays in the receipt of authorisation by WFP in Tanzania, the three months maize ration was not distributed until the 19th Convoy on 2 September and then only on the initiative of TCRS drawing on settlement stocks. The first eighteen convoys, totalling 4,616 persons, had to depend on maize distribution in Mozambique by World Vision International or the Mozambique Red Cross. Given the poor communications, other logistic difficulties with refugees ex-Tanzania scattered among as many as ten different districts and with 133 returnee destinations in Niassa Province alone and thin agency representation on the ground, it would clearly have been preferable to distribute the maize ration in Tanzania from the start.

(239) At the end of the operation in October 1994, a residual caseload remained at Likuyu of 4,502 refugees, most of whom UNHCR expected to repatriate spontaneously without UNHCR assistance after the next harvest between August and October 1995. However, 441 have expressed the desire to remain and be naturalised as Tanzanian citizens. UNHCR is currently planning the closure and handover of the Likuyu settlement to the Tanzanian authorities and is exploring with them and with its NGO partners proposals for the settlement's future use.

(240) Meanwhile, because of the remoteness of so many of the refugees' areas of origin, IOM continues to provide transportation to their homes for spontaneous returnees on arrival in Cabo Delgado Province. Between June 1994 and January 1995, 4,331 persons were assisted in this way, although measures have apparently had to be taken (not surprisingly) to counter the danger of abuse.

SELF EVALUATIONS

(241) One of the most encouraging aspects of the Mozambique repatriation has been the readiness of UNHCR field officers to evaluate their own performance and, through the joint participation of representatives of the country of asylum and the country of origin, to do so in cooperation with each other68.

(242) On the Tanzanian side, a Lessons Learned Survey was conducted in Pemba, Mozambique on 27-28 January 1995 attended by representatives of both UNHCR country offices, and of the principal implementing agencies, notably TAMOFA, TCRS and IOM. The survey was conducted with reference to successive articles of the Tripartite Agreement and the most significant of the lessons learnt were as follows:

1. The operation was greatly facilitated by the use of the VRF, of cross-border travel authorisations and of simplified immigration, customs and health formalities.

2. As far as possible, repatriation operations should involve a few high quality implementing partners, each with clearly defined responsibilities.

3. The PACTOR telecommunications system worked well. Functional mobile radios should be deployed in all vehicles but need immediate qualified technical support in the case of malfunction. This had not been available and there were other purchasing and installation delays.69

4. Many problems arose from the late establishment of field offices in Lichinga (Mozambique) and Songea (Tanzania). These included inadequate assessment of transport routes, delayed registration and too hectic a movement schedule.

5. Implementing partners and operational staff should have been more involved in Tripartite Commission meetings so as to ensure that the recommendations made reflected realities on the ground.

6. Logistic problems included a lack of regional coordination with other simultaneous operations e.g. repatriation from Malawi, so that transport capacity was severely stretched. More detailed information should have been conveyed to implementing partners in the country of origin concerning, for example, the health of returnees and the amount of baggage to be conveyed.

(243) A later report by Sub-Office Songea (made available in draft to the evaluation mission) also emphasised that it should have been established earlier in the year in order to have all the required infrastructure and equipment in place before repatriation activities began. The report also suggested that a visit by the Tanzanian implementing partners to proposed routes and transit centres in Mozambique before the repatriation would have cleared up much of the confusion about the states of routes, bridges and transit centres which bedevilled the operation. The amount of work which needed to be done on several of the selected routes was seriously underestimated and, in the event, they would be used neither for the transportation of returnees nor for the distribution of assistance to them after their arrival.

CONCLUSIONS

(244) All the above are reasonable and sensible conclusions. In general, judged as an exercise in organised repatriation to a strict timetable in difficult logistic circumstances, the Tanzanian operation was an unqualified success. The process was entirely voluntary, the refugees were returned in a safe and dignified manner, no lives were lost, all essential services were provided and there was excellent cooperation between the agencies and authorities concerned. There were only two significant delays in convoy movements and neither of these had adverse repercussions on the operation as a whole. Problems over the availability and accessibility of routes were successfully overcome, despite some wasted effort, particularly on the part of TCRS.

(245) Inevitably, some less positive sentiments and responses can be discerned. On the part of the UNHCR Branch Office for Tanzania, a sense of undue delay at Headquarters over the creation of posts and the re-opening of the Sub-Office at Songea, occasioned partly by Tanzania being accorded a lower priority because of smaller numbers than other countries of asylum and partly because the Burundi crisis of late 1993 and the influx from Rwanda in April 1994 took precedence over the Mozambique repatriation. The Branch Office also felt at a disadvantage in being responsible to Desk V of the Africa Bureau while all the other Branch Offices involved in the repatriation reported to Desk IV. As one staff member put it "we fell into the cracks between the desks" - be that as it may, it is clearly better sense to have the same desk responsible for all aspects of what is essentially or should be a unified operation.

(246) On the part of UNHCR Tanzania and the two implementing NGOs alike there was a sense that forecasting and planning, particularly over routes, was faulty on the Mozambican side, especially on the part of IOM and the road reconstruction NGO, RRR or National Ecumenical Committee. Optimism and assurances gave way too often to last minute denials, necessitating the use of only three routes instead of the projected five with resulting adjustments and delays. It is common ground, of course, that the governmental structure in Mozambique (and perhaps most particularly in this part of Mozambique) was fragile by comparison with Tanzania and that there was not the same level of NGO support. Reception arrangements in Mozambique were often weak and, in retrospect, it was probably fortunate that the majority of the refugees returned by a single route.

(247) On the Mozambican side, a feeling was expressed - not confined to the Tanzanian operation - that the timetable was too rushed, predicated as it was by a desire to complete the repatriation before the elections in Mozambique and the onset of the rains. To the outside evaluator, this impression of too tight a schedule could have been avoided had the UNHCR presence at Songea - and, indeed, at Lichinga and Pemba - been established at the beginning of 1994 rather than, at least in the case of Songea, when the operation was about to begin. In a very real sense, success was the result of careful planning by the Branch Office and by its implementing partners in advance and of the quality and experience of the staff member recruited and assigned to head the Sub-Office at Songea. As almost always it was human attributes that saved the day. We are not always so fortunate.

ZAMBIA

(248) The population of assisted Mozambican refugees in Zambia at 31 December 1992 is given as 25,434 in the Plan of Operation of May 199370. All of these were resident on the Ukwimi agricultural settlement in the Petauke District in Zambia's Eastern Province and many had become relatively prosperous through farming and other income-generating activities. The settlement had been developed and administered since 1987 under the aegis of the Lutheran World Federation, (LWF) one of UNHCR's most experienced implementing partners in the rural settlement field.

(249) Despite the success of Ukwimi71 and the fact that the refugees were under no pressure from the Zambian authorities to leave, it was always believed that the great majority - UNHCR's September 1993 Repatriation and Reintegration Update gives an estimate of 98% - would return to Mozambique. Nearly all the refugees came from the Tete Province, mostly from the Maravia border District and many had ethnic ties on the Zambian side of the border.

(250) A Tripartite Agreement was signed on 15 July 1993 and a Tripartite Commission established. The Commission immediately decided that the main repatriation movement to Maravia should be delayed until 1994, when conditions in Maravia had improved. Meanwhile, an initial movement of some 2,400 refugees who were particularly anxious to return to Chifunde and Chiuta Districts was agreed and took place on 16 October 1993.

(251) The main movement to Maravia began on 8 July 1994 and was completed on 5 December 1994. The LWF was UNHCR's implementing partner throughout. The final figure of those transported under UNHCR/LWF auspices is given as 19,927 in the January 1995 Progress Report72 with 151 electing to remain in Zambia and 50 to return of their own accord in 1995. The Ukwimi settlement is being handed over to the Zambian authorities.

(252) Despite some logistic difficulties, including the crossing by boat of the Luangwa River, the operation appears to have gone smoothly and well. Some 25 convoy movements by bus and truck were organised by LWF and there was also some spontaneous movement from Ukwimi by about a thousand refugees. As in Tanzania, the presence of an experienced implementing partner with a long background of association with the refugees must have proved invaluable and this is a factor to which the Branch Office itself has drawn special attention. The analogy with Tanzania is obvious in other ways as well.

(253) The evaluation mission did not visit Zambia and no particular issues have been brought to its attention. There seems no doubt whatsoever that the operation was well conducted and entirely successful. A minor quibble might be that UNHCR's January 1994 Progress Report says (page 9 para 46 and 47).

"In 1994, the remaining estimated 22,000 refugees will return to Mozambique after the harvest in June ....... The entire repatriation will have to be completed in two to three months so that the refugees can be home in time for the elections and the planting season".

(254) In fact, the entire repatriation took five months and was not completed until after both the elections and the planting season. If there is a lesson to be learnt, it is of the 'don't count your chickens' variety - don't forecast timetables too precisely six months in advance.

ZIMBABWE

HISTORY AND BACKGROUND

(255) Following some earlier arrivals due to drought and famine, Mozambican refugees began to arrive in Zimbabwe in considerable numbers from 1984 onwards. The border between Zimbabwe and Mozambique stretches from the Tete Province in the north via the Manica Province to the Gaza Province in the south and the majority of refugees came from these provinces with smaller numbers from Sofala, Zambezia and elsewhere. Some settled spontaneously, some were put immediately into camps and, particularly in the early stages, some who had settled spontaneously were moved by the Zimbabwean authorities into camps.

(256) By the end of 1992, there were five refugee camps spread from north to south at distances from the border ranging from 80 to 120 kilometres. The distances from the camps to border crossing points into Mozambique were often a great deal longer than this, however, while the spontaneously settled might be anywhere in Zimbabwe. The camps - Nyamatikiti, Mazowe, Nyangombe, Tongogara and Chambuta - were between 240 and 580 kilometres from the Zimbabwean capital, Harare. All dated from 1984, except Chambuta which was not opened until 1991.

(257) UNHCR started to assist the camp population of Mozambican refugees in 1985 but had only limited access to and no presence in the camps which were administered by the Government with the support of NGOs. By the end of 1992, the official camp population was 136,641 but had risen to 137,848 by the end of February 1993. Most of the refugees were of an illiterate, subsistence farming background but, since the camps were fenced and only vegetable gardens were allowed, were supported almost entirely under a conventional care and maintenance programme.

(258) A Tripartite Agreement for the repatriation operation - the first since the Malawi/Mozambique agreement in 1988 - was signed on 22 March 1993. As has been noted elsewhere, it was a considerable disappointment to the Branch Office that the Agreement, as finally approved by Headquarters and accepted by the signatory governments, followed a standard model and did not include more specific provisions covering, for example, the scope of the information campaign, registration policy, vulnerable groups and coordination with other Tripartite Commissions.

(259) The Plan of Operation of May 1993 envisaged that some 30,000 of the camp population of refugees would return home by certified self-repatriation with a cash travel grant (estimated by the Branch Office at about US$27 per head) with the remainder of up to 108,000 returning under organised repatriation arrangements with transportation provided by UNHCR73. Organised movements were not expected to start until the middle of 1993, with the bulk taking place in 1994. There had been large-scale destruction in the areas in Tete and Manica Provinces from where up to 70% of the refugees originated and Zimbabwe had the unique problem among all countries of asylum that both sides of the border had been mined, the Zimbabwean side under the Rhodesian/UDI regime.

(260) After some initial enthusiasm, however, the Government of Zimbabwe (wisely, in the opinion of this evaluator) decided against certified self repatriation on the eminently sensible grounds that some might spend their repatriation entitlement and, having 'blown' their grant, might never repatriate at all. The Government insisted that all repatriation (other than the purely spontaneous) should be by means of UNHCR-organised transport so as to ensure that returnees actually crossed the border into Mozambique and did not settle elsewhere in Zimbabwe. This remained its stance throughout the operation and, at a meeting of UNHCR Representatives from Southern Africa held in Geneva between 29 March and 1 April 1993, it was agreed that the cash grant proposal for Zimbabwe should be dropped. This decision was not, however, reflected in the May 1993 Plan of Operation which maintained the reference to certified self-repatriation (para.197)

PLANNING

(261) By March 1993, Branch Office Zimbabwe had compiled - there were at least five successive drafts - a comprehensive country Plan of Operation covering

(1) The pre-departure phase, comprising an information and promotion campaign, a Mine Awareness Campaign and registration for repatriation;

(2) The movement phase, to be carefully programmed to reflect the absorptive capacity of home areas in Mozambique.

(262) The vast majority of the refugees were expected to return and the operation was expected to last until early 1995.

(263) The country plan was extremely thorough and detailed and is, in fact, a model of its kind. It dealt at length with registration procedures and computerization, health, education, the needs of vulnerable groups, transportation and communications. A key element, which was wisely adhered to throughout the operation, was that the assisted repatriation of spontaneously settled refugees would only be considered after organised repatriation from the camps was at an end. It was realised that UNHCR and its implementing partners did not have the capacity to undertake both operations simultaneously and needed to gain experience of organised camp repatriation first before undertaking the even more complex task of identifying, assembling and repatriating the spontaneously settled. The number of spontaneously settled refugees was estimated throughout as 100,000, a figure which probably owes as much to guesswork and a fondness for round numbers as to any profound statistical or demographic consideration.

(264) Staffing requirements had already been notified to Headquarters on 8 February 1993, following a mission by the Coordinator, the Head of Organisation and Methods and the Senior Administrative Officer from the Regional Bureau for Africa. The key post of Durable Solutions Officer was already filled, but it was not until April 1994 that there was a UNHCR team led by an Associate Field Officer in each of the five refugee camps. Two Field Officers were appointed, one in September and one in October 1993 but other programme and logistics staff did not arrive till early 1994. The fact that National Officer appointments had to be made by the APPB caused considerable delays.

(265) The only respect in which the Country Operation Plan went seriously awry was in making the assumption (5th draft page 32) that about 80% or 110,000 would return by certified self-repatriation starting in April 1993 with the remaining 30,000 travelling by UNHCR organised transport74. As we have seen, the self-repatriation concept was abandoned shortly thereafter and UNHCR had to arrange convoy transport for the entire caseload.

(266) The document remains a most impressive achievement. It was further supported by a compilation of planning data for voluntary repatriation produced by the UNHCR Branch Office in May 1993 on the basis of information provided by the Government's Department of Social Welfare (DSW) and by NGOs working in the refugee camps (in which, it must be remembered, UNHCR had as yet no established presence). This contained, in more detail than is available for any other country of asylum, particulars and figures by camp of the refugees' districts of origin in Mozambique, the numbers in various vulnerable groups, information on unaccompanied minors and a variety of skills profiles. All this data was communicated to UNHCR Maputo even though the situation in Mozambique was not conducive to its being fully used in the reception and reintegration process.

(267) A crucial second meeting of the Tripartite Commission - the first seems to have been a formality - was held in Harare between 2 and 4 June 1993. The Coordinator for Southern Africa was present from Geneva and both Zimbabwean and Mozambican delegations were led at ministerial level. At this meeting, preparations for which had been made thoroughly in advance, the UNHCR delegation sought to supplement the general terms of the Tripartite Agreement with more detailed provisions75.

(268) These covered, inter alia, border crossing points and procedures, luggage entitlements, immigration and customs formalities, registration and the information campaign. It was agreed that organised repatriation from the camps would comprise (I) a limited programme between July 1993 and April 1994 of movement to 'target' areas which the Mozambican authorities and UNHCR Mozambique had confirmed were in a position to receive returning refugees, and (ii) a larger scale operation between April 1994 and April 1996.

(269) Several important decisions were taken, all of which contributed to the success of the operation. It was agreed that immigration and customs formalities would be concluded in the camps prior to departure, so as to avoid lengthy border checks. Registration would be done approximately two months in advance of the movement of a particular group chosen on the basis of absorptive capacity in Mozambique. Repatriation would take place from all camps in parallel rather than in sequence emptying one camp then moving to another. This would be more acceptable politically and less threatening to camp employees whose numbers could be scaled down as the care and maintenance programme reduced in size.

IMPLEMENTATION

(270) The voluntary repatriation of refugees from the five refugee camps began on 12 June 1993 and ended in December 1994, apart from one convoy from Mazowe River Bridge camp on 7 March 199576. Only a small residual caseload remains and a ceremonial handover of the camps to the Government of Zimbabwe took place on 20 December 1994.

(271) Refugees were repatriated by road convoy to no less than 63 reception centres in Mozambique, travelling distances of up to 800 kilometres before reaching their destination. The majority of the refugees came from either Tete, Manica, Sofala or Gaza provinces with a minority from further afield. Refugees were only registered for and moved to areas where UNHCR Mozambique had indicated that they could be received and any ceiling figures given were strictly adhered to. The absorptive capacity in the country of origin was a prime determinant in the planning of convoy schedules. The focus of cooperation was direct liaison between UNHCR Harare and the UNHCR Sub- and Field Offices in Mozambique where repatriants would be received. Regular repatriation coordination meetings were held between the two sides and day-to-day operational issues resolved between field staff direct. As elsewhere, decentralised field relations between country of asylum and country of origin were a key element of success.

(272) Once UNHCR Harare had been informed by UNHCR Mozambique (usually the relevant Sub-Office) that repatriation to a particular area could begin, registration for the area commenced. Registration was carried out by teams consisting of project staff employed by UNHCR's implementing agency, Inter-Country People's Aid (IPA), a locally-based NGO. Up to the end of 1994, each of the five camps had a Registration/Repatriation team of twelve project personnel, reduced to three mobile teams for the repatriation of the spontaneously settled in 1995. In 1994, a qualified social worker was added to each team specifically to concentrate on the needs of vulnerable groups.

(273) Registration was preceded by a comprehensive information campaign in the form of leaflets in English, Portuguese and Shona, a locally produced video, radio broadcasts, television announcements and meetings with refugees. The registration itself was based on the use of the Voluntary Repatriation Form (VRF), a procedure which UNHCR Zimbabwe took more seriously than did UNHCR staff in some other countries of asylum and would, indeed, like to have seen more geared to reintegration needs in the country of origin77. The information entered on the VRFs was computerised in the field by EDP Assistants, who formed part of the Registration/Repatriation Teams, an approach which the Branch Office felt worked much better than the centralised computerisation of VRF data at Branch office level which had been a feature of earlier repatriation operations e.g. Somalia to Ethiopia in 1990.

(274) The first repatriation convoy to Mozambique took place on 12 June 1993 in response to a UNHCR Headquarters' request dated 27 May - some months before it had originally been intended that movements would begin. No Field Officers were yet in place, no Registration Teams had yet been appointed and the VRFs had not yet arrived. Nevertheless, a convoy of 254 persons was assembled from Nyangombe camp, the first of 19 which transported 9,316 refugees up to the end of 1993. By the end of 1994, the total number of convoys had risen to 151 and 92,069 refugees had been repatriated. With a final convoy of 1,098 persons from Mazowe River Bridge camp in March 1995, the total camp population repatriated stood at 93,167 for the entire operation. This was not far short of the cumulative target for 1993 and 1994 of 100,000 (10,000 for 1993 and 90,000 for 1994) bearing in mind that there had also been spontaneous repatriation from the camps and that the verified camp population was only 96,380 at the end of 1993, clear evidence that earlier figures had been inflated.

(275) For each convoy, there was a comprehensive checklist for the guidance of field staff covering 53 listed items, a further checklist of 24 items for the Branch Office repatriation staff and a final list of 19 items to be checked whenever unaccompanied minors were being repatriated. An impressive feature of the operation was the particular attention given to the special problems of vulnerable individuals and groups, such as the elderly, the sick and handicapped, unaccompanied women and the heads of single parent households.

(276) NGOs working in the camps played a prominent part - for example, the Save the Children Federation USA (SCF-USA) was responsible for seeking durable solutions for unaccompanied minors and for ensuring that no unaccompanied minor was returned to Mozambique unless family members had been traced or suitable foster or institutional arrangements could be made. Inter-Country Peoples Aid (IPA) conducted a health training programme prior to repatriation concentrating on infant feeding, the prevention of AIDS and the treatment of diarrhoea78.

MINE AWARENESS

(277) IPA also conducted a Mine Awareness Training Campaign (MATC) funded by UNHCR for all Mozambican refugees in camps (it was obviously impossible to offer systematic training of this kind to the spontaneously settled who repatriated in 1995). The programme was actually initiated under the auspices of an international NGO Help Age (UK) in mid-1993 and taken over by IPA on its foundation as an indigenous Zimbabwean NGO in January 199479.

(278) Mine Awareness activities began in August 1993 with workshops for MATC on course syllabus content and methodology conducted by experts from Norwegian People's Aid. In September and October, refugee instructors were trained in the camps and from October 1993 onwards courses were offered which by the end of the programme had trained 79,979 refugees. This total included children; in January and February 1994, primary school teachers were trained so as to introduce Mine Awareness as part of the syllabus in camp schools.

(279) All the mine awareness training materials - posters, banners, videos, models and T-shirts as well as a manual in Shona and English - were produced locally and copies of materials and advice on training were provided to other countries in the region such as Malawi and Zambia. However, the video was actually 'shot' by a video crew from Headquarters in Geneva. Produced in three different languages - Shangaan, Nyungwe and Shona and with three casts it proved a great attraction in the camps and helped to break down apathy towards the subject. Nevertheless, all those concerned with the programme in Zimbabwe believe that it could have been produced as well or better locally and point out that Zimbabwe and southern Africa generally have ample resources for video and film work.

(280) They also believe that the start of the programme was delayed because of a desire to centralise in Headquarters the production of training materials for the entire region with the support of a mine awareness consultant based in Maputo, whose contribution to the development of the programme remains unclear. At any event, materials were apparently not forthcoming from Geneva and the local organisers went ahead without them.

(281) One unique feature of the programme was the association with it throughout of military personnel in the form of sixteen engineers from the Engineering Directorate of the Zimbabwe National Army. These became - in plain clothes - an integral part of the MATC teams, made an outstanding contribution to the programme and were proud to be associated with it.

(282) The refugee trainers themselves were under the direct supervision of the engineers in each camp. Their own motivation was enhanced by the issue of uniforms, shoes and other material incentives as well as Certificates and Letters of Appreciation at the conclusion of their work. They also received training in First Aid from the Zimbabwe Red Cross.

(283) The Mine Awareness training programme in Zimbabwe has been described as some length because, without in any way detracting from the programmes in other countries of asylum it seems a model of its kind and one which would be capable of transfer as a 'package' with the same core management and personnel to similar situations elsewhere in Africa. Particularly striking features of the programme were the use of seconded military personnel and the emphasis placed on teaching mine awareness to children in camp schools. The organisers were conscious throughout that mines were a danger on both sides of the Zimbabwe/Mozambique border with minefields having been laid extensively by the Rhodesian Army during the independence struggle.

TRANSPORT

(284) As we have seen, repatriation from Zimbabwe to Mozambique involved road travel over exceptionally long distances of up to 825 kilometres from camps to border crossing points (Nyamatikiti to Chicualacuala is the extreme case) and, although the roads in Zimbabwe were good, those within Mozambique to reception centres or even final destination were often extremely bad80. Although some destinations could be reached by buses, many could not and a fleet of four-wheel drive trucks and trailers became a necessity.

(285) Various possibilities were explored. Contrary to popular belief, the hiring of commercial transport did not prove easy, only 12 transport operators showed an interest and all save two were excluded because their fleet was unsuitable or too old or withdrew themselves because of problems with spare parts, security fears and other reasons. Charges were inflated in the common belief among transporters that UN agencies will pay the earth, including danger money in hard currency for venturing into Mozambique. The two companies which remained on offer - TENDA and KUKURA KURERWA were contracted to provide buses and not trucks.

(286) Alternative solutions also failed. Transporters from other countries are prohibited from operating in Zimbabwe and, even if allowed, would have had to pay heavy border fees and taxes for each entry into Mozambique. The use of vehicles from the Zimbabwe National Army was not acceptable to the Government of Mozambique and NGO capacity in Zimbabwe was insufficient for more than supplementary support. The decision taken, therefore, was to procure a UNHCR fleet of trucks and trailers and to hire buses from local transporters for moving repatriants along routes which buses could reach.

(287) From 1 October 1993, the transport logistics operation, including the management and servicing of the UNHCR transport fleet, was entrusted to the German Technical Cooperation organisation (GTZ) under a Tripartite Agreement between GTZ, UNHCR and the Government's Department of Social Welfare. At its maximum size, the repatriation fleet consisted of 43 Mercedes Benz 4WD trucks, 28 trailers, 7 Nissan trucks (transferred in good condition (!) from the Cambodia repatriation and a most useful acquisition), 3 Volvo trucks, 3 Mobile Workshops and 5 Escort vehicles. An average of 30 buses were on hire at any one time. GTZ appears to have performed outstandingly well and to have been so satisfied with its relationship with UNHCR that, at an interview with the evaluation mission on 3 May, the GTZ Project Manager for Zimbabwe declared his organisation ready to cooperate if needed in similar ventures in the future. So many NGO/UN relationships seem to turn sour that it is a pleasure to hear of one which remained sweet throughout - and a credit to all concerned.

(288) Provided the vehicle fleet can be satisfactorily and economically disposed of at the end of operations - best of all, transferred as a viable entity to another programme - this seems an entirely acceptable approach to transport management, provided a suitable NGO partner is available to administer the fleet. UNHCR should, under no circumstances, try to do so itself; it may hire as many so-called logistics officers as it likes but, in the last analysis, lacks the skills and aptitude to be transport manager on its own account.

(289) A cost comparison with other country of asylum operations would be misleading, since conditions differed so much from one country to the next. But the total Zimbabwe repatriation budget for the 3-year period 1993-95 was just under US$11 million and this was not exceeded. The Branch Office estimates the total cost per repatriant as about US$60.

TELECOMMUNICATIONS

(290) Because there was no permanent field presence in Zimbabwe before the repatriation operation, a regional telecommunications network had to be built from scratch. Equipment arriving in bits and pieces with a high turnover of visiting radio technicians meant that effective functioning took a long time to achieve. The eventual assignment of one technician to service Zimbabwe's needs on a permanent basis was a great help, particularly in view of the excellent quality of his work, but the fact that he was based in Maputo and had to cover South Africa as well was a persistent handicap. Training was neglected, even though there were training personnel in Maputo.

(291) Operational needs would have justified a full-time technician stationed in Harare, at least at the outset and when activity was at its peak. Expensive and vitally necessary equipment is involved in large scale operations of this kind and cannot be properly installed and maintained if it is haphazardly delivered and put in place in a peripatetic fashion.

REPATRIATION OF SPONTANEOUSLY SETTLED REFUGEES

(292) The initial approach to the repatriation of spontaneously settled refugees was a cautious one. The country Plan of Operation (page.10 para.3.2.1) identified them as a second target group but proposed that the policy towards their registration for repatriation should be determined while the registration of the first target group - those in camps - was going on. The May 1993 Plan of Operation (page 41 para.194) said of the spontaneously settled that the possibility of their being incorporated into the programme would be considered at a later stage but added an assumption that they would not require assistance to return. Their numbers were consistently stated as about 100,000 but this was clearly the vaguest of estimates and emanated from government sources. No reliable statistics have ever been or are likely to be compiled by UNHCR, the Government of Zimbabwe or anyone else. The operational objective was to assist 20,000 to repatriate and this was more than fully achieved.

(293) Despite the general atmosphere of caution, the UNHCR Branch Office in Harare has always accepted that Mozambicans spontaneously settled in Zimbabwe are eligible to be considered as refugees even though it had nothing whatsoever to do with them prior to the repatriation operation. This 'prima facie' acceptance was a historical legacy which those dealing with repatriation had to face. It also reflected policies and practice of the Government of Zimbabwe which recognized spontaneously settled Mozambicans as refugees and wished their voluntary repatriation to form part of the UNHCR programme.

(294) This recognition may, or may not, have had a basis in law. The Zimbabwe Refugees Act No.13 of 1983 provides in Section 3(1)(e) that a person shall be a refugee for the purposes of the Act "if he is a member of a class of persons declared in terms of subsection (2) to be refugees". Subsection (2) provides that if the Minister considers any class of persons to be refugees within the definition of the Act he may declare them to be refugees. Subsection (3) provides that any such declaration shall be published in the Gazette.

(295) The Branch Office, Harare understands that a declaration has been made by the Minister to the effect that all Mozambicans in Zimbabwe are refugees. But the Branch Office has never been able to find it in published form. To find such a declaration in the official Gazette should be a relatively simple matter and the Commissioner for Refugees ought, in any event, to be able to provide the appropriate reference. If it has not been published, then it has not been properly made and is not a legal declaration. In the opinion of this evaluator, it has probably not been made in the appropriate form, although it may well be the Government's wish that all Mozambicans on Zimbabwean territory be considered refugees and eligible for repatriation under UNHCR auspices.

(296) In any event - as in South Africa - there was really no choice if good relations with the Government were to be preserved and if genuine refugees who had settled in Zimbabwe, often in penurious circumstances, were to be assisted to return home in conditions of dignity and safety.

(297) Nevertheless, it had to be accepted that there was no feasible way of distinguishing mandate refugees from persons who UNHCR, if not the Government of Zimbabwe, would properly regard as economic migrants. But many who had arrived spontaneously in Zimbabwe in the early 1980s had been moved by the authorities into camps, while a good number in the camps had themselves elected to leave and return spontaneously to Mozambique81 - which helps to account for the fact that UNHCR sponsored organised repatriation from the camps only moved 92,069 refugees between mid-1993 and the end of 1994 out of a camp population, according to government figures, of nearly 140,000.

(298) Moreover, a number of the spontaneously settled were believed to have returned to Mozambique on their own initiative before the operation to assist them began. When in January and February 1994, the Branch Office in conjunction with the Zimbabwean Department of Social Welfare carried out a data collection exercise and information campaign in areas where there were likely to be large concentrations of the spontaneously settled, only some 12,000 Mozambicans were identified as wishing to repatriate, a far cry from the original estimate of 100,000, most of whom were expected to return.

(299) Furthermore, it is reassuring that there was no indication, in contrast to South Africa, of a 'revolving door' syndrome and neither UNHCR nor the Government was aware of any reverse flow of returnees back to Zimbabwe from Mozambique82. This would suggest that those returning intended, at the very least, to return home to participate in the rebuilding of their society - probably more than half came from the neighbouring Mozambican province of Manica and could easily have filtered back into Zimbabwe had they wished to do so. Many of the spontaneously settled were in border areas earning a precarious living as underpaid workers on commercial farms and the farmers had, by and large, no desire to see them leave. There were allegations during the operation that some had been forcibly prevented from leaving by their employers or that their wages had been withheld to induce them to stay. Relations between the Commercial Farmers Union and UNHCR became strained and there were threats that the tobacco harvest would suffer from a shortage of labour if too many Mozambicans left. None of this suggests any pressure or compulsion on the spontaneously settled to return to Mozambique.

(300) But there is another side to the story. UNHCR officials in Mozambique - particularly the Heads of the Sub-Offices in Chimoio and Tete, from which Provinces the great majority of Mozambicans in Zimbabwe originated - had always been anxious that the operation to repatriate the spontaneously settled should not be of prolonged duration, as the reception and registration network in Mozambique was due to be dismantled and energies concentrated on reintegration.

(301) They were also frankly sceptical about the voluntary nature of the repatriation and even about the desire of Mozambicans left in Zimbabwe to go home. In a memorandum dated 23 February 1995 from the UNHCR Representative in Maputo to his counterpart in Harare, he said:

"From the information received lately it would seem that we may be moving into the direction of doing a survey (or even a census) of Mozambicans (i.e. not refugees) still remaining in Zimbabwe, the great majority of whom seem to be migrant workers. Furthermore, it seems clear from the mission report that there is very little interest in this operation on the part of the potential beneficiaries, their employers and even the local authorities".

(302) The mission report he referred to had been prepared by an Associate Logistics Officer from the Branch Office at Harare, suggesting an increased timeframe for the operation over and above the two months originally envisaged, and had queried the voluntary nature of the operation in the sense that word had been spread that the Government of Zimbabwe wanted to see all Mozambicans return home and might in the future deport all those who were not legally resident in Zimbabwe, under the provisions of the Immigration Act. The report had sufficiently alarmed the Head of the Chimoio Sub-Office that, supported by the Head of the Tete Sub-Office, he had asked for a thorough discussion with Branch Office, Harare. This prompted the Mozambique Representative's memorandum to Harare in which he also suggested that 'possibly, not more than 2,000 persons will opt to repatriate', an estimate which turned out to be well short of the mark.

(303) Fortunately, relations between the two Branch Offices and the concerned Sub-Offices were close and cooperative and had been so from the start of the operation. The Mozambique Representative was able to conclude his memorandum with a reassurance that his counterpart in Harare shared the same concerns and that the operation would finish by the end of May 1995. The issues continued to be discussed between UNHCR Mozambique and UNHCR Zimbabwe at coordination meetings, in particular on 24 February in Harare and on 4 April 1995 at Chimoio, where the voluntary nature of the operation was again stressed by BO Harare and a number of operational issues resolved.

(304) Nevertheless, it would be true to say that UNHCR Mozambique's doubts about the voluntary nature of at least some individual movements were never wholly allayed. Some returnees continued to allege on arrival in Mozambique that they had been forced to return or told that they would be put in jail if they did not do so. Others seemed bewildered and single young men (on one occasion in the presence of the evaluation officer) were apt to invent stories which they thought might be to their advantage.

(305) Yet, on the whole, the operation went extremely well. The movement phase started on 14 March 1995 and by the end of April almost 12,000 had been successfully repatriated. This was in spite of a temporary suspension of several weeks in April because of the General Election and the Easter holidays immediately thereafter, when Government counterparts would not have been available. The suspension also allowed tobacco harvesting to be completed, following a request by the Commercial Farmers' Union and the Tobacco Association to the Government's Department of Social Welfare (DSW). By universal agreement, and with some relief, the operation was concluded at the end of May with a final repatriation figure of 24,535.

(306) How many returned to Mozambique entirely of their own accord over the entire duration of the repatriation exercise between 1993 and 1995 is, of course, anybody's guess. In February 1995, the Branch Office in Maputo quoted figures, presumably from Government of Mozambique sources, of a cumulative total of 192,439 returnees to Manica Province and 105,516 to Sofala Province83, numbers far in excess of the entire camp population in Zimbabwe and before the UNHCR-assisted return of the spontaneously settled had even begun. What credence can be given to these figures is hard to say (there are similar ones for returnees from South Africa), but if they are anywhere near the truth, they suggest that by far the greater part of the Mozambican refugee population in Zimbabwe repatriated of its own accord.

(307) The repatriation movement of the spontaneously settled was organised in much the same way as the movement from the camps, with the obvious exception that registration could not be carried out in advance and the nature and timing of the operation did not permit the mounting of a Mine Awareness campaign. A massive information campaign was carried out by means of pamphlets, posters, newspaper advertisements, radio and television broadcasts and meetings under the auspices of three joint UNHCR/DSW mobile teams. Refugees wishing to repatriate were told to make their way to pre-determined departure points from where they would be divided into groups according to destination and taken by road convoy to one or other of seven selected reception centres in Mozambique and onward transportation thereafter to their homes. No infrastructure was put in place at the chosen departure points, which were nothing more than open spaces, football stadia and the like. Wherever possible, the UNHCR Zimbabwe buses and trucks took the refugees on to their final destination.

(308) In a position paper on the repatriation of the spontaneously settled forwarded to UNHCR Headquarters on 22 April 1994, the Branch Office at Harare commented that 'logistical and other operational constraints ... will constitute a predictable nightmare'. This Cassandra - like prophecy proved, in the event, to be far too pessimistic and 'the constraints', which were certainly formidable, were remarkably well overcome. This was due to firstly, thorough advance planning of which this excellent position paper, drafted almost a year ahead of the actual commencement of this phase of the operation, is a prime example. Secondly, it was due to the long experience already gained in the repatriation from the camps and the close cooperative relationship which had been established between UNHCR Zimbabwe and UNHCR Mozambique and with the authorities on both sides of the border. Thirdly, it was due to first class organisation on the ground, which involved far more than some at Headquarters, who had apparently envisaged the operation as a symbolic gesture, can ever have envisaged.

(309) An invaluable part of the evaluation mission was a visit which the Branch Office in Harare had organised to witness the preparations for and the departure of Convoy No. 178 (!) from Guruve in Mashonaland Central Province on 5/6 May 1995. This made it possible to witness at first hand all the arrangements for a smooth and orderly movement which an outstanding team of international, national and NGO partner staff had made (not forgetting the excellent cooperation of the Zimbabwean authorities).

(310) The refugees assembled in an open space at Guruve during the day and well into the evening. They were registered, a convoy manifest prepared - all computerised - and immigration formalities completed by Government officials on the spot. An evening meal was cooked, food and water, with an issue of jerry cans per family, supplied and the refugees split into groups according to destination. During the night, buses, trucks for baggage, and Zimbabwean police escorts arrived and within an hour from about 3am to 4am, the buses and trucks were filled and the different parts of the convoy moved off to their various destinations in Mozambique, some with many hundreds of kilometres to travel. Everything was conducted in the most efficient, humane and good-humoured manner possible, a remarkable achievement which reflects great credit on all concerned and encouraging evidence that, in management terms, UNHCR can get it right.

CONCLUSIONS

(311) The voluntary repatriation from Zimbabwe of a grand total of 117,702 refugees was thoroughly and meticulously planned and executed with great efficiency and precision. Just about every foreseeable eventuality was foreseen and accounted for in the planning process. The Durable Solutions Officer had an above average competence in computerised aids to planning and considerable experience of repatriation operations elsewhere in Africa. But the foundation of this operation's undoubted success was laid by her and the Representative's careful planning in advance.

(312) For example, on 18 December 1992, almost exactly six months before actual repatriation movements began, the Branch Office sent the Coordinator at Headquarters a suggested division of responsibilities between UNHCR, the Government, other UN Agencies and NGOs which had already been agreed with the Government's Commissioner for Refugees. This was followed by the detailed Plan of Operation, a division of responsibilities at camp level and the various check lists for implementing staff which have already been mentioned. Documents such as these and various compilations of planning data, information pamphlets and the like formed the basis for the implementation of the programme. Careful planning and distribution of duties also made it possible to pay particular attention to the needs of vulnerable groups.

(313) Another commendable aspect of the Zimbabwe operation has been the emphasis laid throughout on close cooperation with UNHCR colleagues in the country of origin and on the absorptive capacity of different areas in Mozambique as the prime determinent of movement schedules. The principle adopted has always been that the reception and reintegration needs of the country of origin should be paramount and should always take precedence over the urge to meet repatriation targets and close down camps. At a certain point the initiative has to pass to the country of origin and the Branch Office in the country of asylum must recognise the fact.

(314) One is left with an uneasy feeling, however, that perhaps this principle was carried a little far in the 1995 repatriation of the spontaneously settled. The span allotted in January and February for data gathering and an information campaign was desperately short given the immense area in Mashonaland Central, Manicaland and Masvingo Provinces that had to be covered. A bare two and a half months was allowed for the actual repatriation - in contrast to sixteen months for South Africa, the only comparable operation84.

(315) The February 1995 report by the Associate Logistics Officer85 (quoted earlier) on his mission to the three field teams preparing for their phase of this operation commented that the time allowed for the information campaign was "totally inadequate and needs to be extended if the programme is to succeed". He went on to suggest - this is the main thrust of the report - that if the movement phase were to be delayed until the middle of May 1995 and extended to September to allow the spontaneously settled to harvest their own crops in the communal areas "the majority of the spontaneously settled Mozambicans will opt to repatriate" (para.5.1 page 6).

(316) As we have seen, this report caused some consternation among UNHCR staff in Mozambique who were phasing out their repatriation and registration teams and wanted to concentrate on reintegration. They were quickly reassured that the operation would end in May. Nevertheless, one is left asking - if there was a genuine desire to offer assisted repatriation to as many as possible of the spontaneously settled Mozambicans, was not the time-frame impossibly short? Admittedly, the operation was not a comfortable one, any more than in South Africa, and perhaps even less.

(317) Attitudes in Zimbabwe towards the operation seemed to vary between apathy, downright hostility by commercial farmers or, in some areas, opposition politicians and near-compulsion to repatriate. It must have been a relief to bring the operation to an end, yet one is left wondering how many more Mozambican refugees might have gone home if they had longer to prepare. The Branch Office's own (excellent) submission to the evaluation mission spoke of the need for 'phased registration which can only take place after an adequate information campaign'. Was either the information or registration/movement schedule adequate in this case?

(318) The Branch Office, Harare held a 'lessons learnt' session between 2-4 June 1995 with colleagues from Mozambique but the conclusions were not yet available as this report was being prepared. Meanwhile, a 'lessons learnt' paper on 4 May 1995 provided by the Branch Office to the evaluation mission covers a number of the points mentioned in this Chapter. The salient conclusions of the paper may be summarised as follows:

(a) the pace of the operation must be determined by the absorptive capacity and material conditions in the country of origin.

(b) a detailed country profile of the country of origin should be maintained in the country of asylum, covering the full administrative structure and incorporating good maps. Refugees should be given detailed information about the conditions to which they would return.

(c) The Tripartite Agreement should be more than just a statement of intent but should incorporate decisions on major policy issues and agreed operational modalities.

(d) The VRF should be designed in consultation with the country of origin and other countries of asylum and computer convenience should not be allowed to dictate the contents. Actual computerisation of registration inputs should be carried out in the field.

(e) Telecommunications systems should be put in place before the start of operations and regularly maintained.

(f) The division of responsibilities between government, UNHCR and NGOs should be agreed in writing during the planning stage.

(g) FMIS software should be shared with implementing partners to enable them to budget, account and report in the required UNHCR format.

(h) Greater emphasis should be placed on the use of local resources rather than on expensive and time-consuming project inputs from elsewhere (Mine Awareness materials, for example).

(i) Cross Border Travel Authorisations have proved of enormous value in facilitating movement.

(j) the staffing needs of new operations warrant a speedy and flexible response.

All sensible suggestions, many of which echo experience in other countries of asylum.

N.B. Since this chapter was drafted, the report on the 2/4 June 'Lessons Learned Workshop' has been received. As might be expected, it is an impressive and comprehensive document of 48 pages and annexes, which should be required reading for those engaged in repatriation programmes in the future. Happily, its conclusions are largely reflected in the previous pages, which there seems no reason to modify or alter.

---

1 Cable 0170 to Headquarters dated 22.2.95.

2 Telex ZIM/HCR/0052 to Headquarters - a copy provided to evaluation mission undated but certainly late January or early February 1993.

3 Fax ZIM/HCR/0530 of 16 December 1992.

4 The Tanzania Tripartite Agreement of 10 November 1993 did include one country-specific clause (Article 14) to cover the conversion of local currency, while being in other respects identical to the specimen text in the Plan of Operation.

5 At the Oslo Conference on 22 June 1992, the Deputy Director of the Regional Bureau for Africa announced the High Commissioner's decision to create such a post to oversee all aspects of the operation, notably planning, implementation, monitoring and coordination. Nothing was said about authority.

6 Undated - the curse of UNHCR documentation through the ages.

7 Repatriation under Conflict, A review of the Encashment Programme for Afghan Refugees in Pakistan. UNHCR/EVAL/ENC/14. February 1994.

8 The DHC objected strongly to the use of travel grants. DHC's Mission Report of 18 June 1993 on mission to Mozambique 6-12 June (para.12)

9 Various academic studies, consultancy and logistic inputs also contributed to the planning process.

10 The DHC's mission report of 18 June 1993 following a mission to Mozambique between 6-12 June strongly recommended that the Coordinator should have line authority over the Desks and with an operational repatriation unit in support (para.24). No action appears to have been taken to implement this view.

11 A full Information Bulletin on Mozambique repatriation was issued virtually simultaneously on 12 May 1993. There was another more general Bulletin on Repatriation in Africa in October 1993. The sections on Rwanda and Burundi make sorry reading today.

12 Some of which, the absorption capacity of NGOs, lack of staff, mined roads, difficult terrain, insufficient transport were highlighted in the October 1993 Information Bulletin "Repatriation in Africa".

13 The writer recalls donor missions organised by the Public Information Section in the 1980s, but not by the Fund Raising Service.

14 almost US$20 million towards 1995/6 needs had already been contributed by 30 June 1995.

15 The Quelimane Sub-Office assumed responsibility for UNHCR operations in Zambezia, Niassa, Cabo Delgado and Nampula Provinces, leaving Sub-Office Tete to deal with Tete Province alone.

16 South Morrumbala District had already received some 69,000 returnees by the time a field presence there was established.

17 While this evaluation is concerned only with repatriation, reintegration activities had to proceed simultaneously with repatriation or, ideally, to precede it to ensure that basic services were provided for those returning.

18 Despite the fact that the programme was directly or indirectly responsible for more than 600 vehicles by early 1995, no Assets Management Unit has been established to deal with these and other physical assets.

19 Including so-called "assisted spontaneous" repatriation

20 BO Mozambique claims that the only significant transfer of resources was of four vehicles from Malawi.

21 The evaluation report does not disclose what this acronym stands for.

22 The movement to Malawi continued even after the Peace Accord of October 1992.

23 "There are no refugees in Malawi and no refugees from Malawi anywhere else" used to be the official line.

24 This cross-border movement became a serious problem in the mass repatriation phase with, as we shall see, those who had already repatriated returning to Malawi to be fed.

25 Field Operations Review, August 1992, report by Joseph B. Stern Annex III Malawi Section 3.0 page 43.

26 The Local Impact of Mozambican Refugees in Malawi by Lynellyn Long, Lois Cecsarini and James Martin, November 10, 1990.

27 Zikomo Malawi (information paper) update No 1. May 9, 1994

28 For example, in November 1993 a verification exercise showed that, out of a feeding figure of 18,000 at Changambika camp in Chikwawa district, only 9,800 were actually living in the camp, the rest having returned to Mozambique. The border was only 5kms from the camp. Food consolidation was implemented at the end of February 1994 by which time a resident caseload of only 2,418 remained.

29 See, for example, the Note on the UNHCR Programme in Malawi prepared for the Regional Representatives Meeting, 6-7 March 1995, in Johannesburg, page 1 para I2.

30 But the UNHCR Mozambique PI Section Special Report "Back on Track" of February 1995 gives a figure of 144,221 assisted returnees from Malawi. Such discrepancies are all too familiar.

31 e.g. 600 through Tengani on 24 November 1994 as against 200 originally agreed (although the terms of this agreement were disputed by BO Malawi)

32 these fears, while understandable, were unfounded as no mass movement did, in fact, materialise. This was confirmed by BO Mozambique at a joint meeting with BO Malawi and WFP on 27 March 1995.

33 UNHCR Mozambique Progress Report of 27 April 1995.

34 The Malawi National Coordinating Committee met eleven times between June 1990 and March 1994. UNHCR and other agencies were represented on both National and District Committees

35 well, nearly entire, one old man didn't want to go

36 "Towards Satisfying Mozambican Refugees' Information Needs in Malawi" Vince Owen, Save the Children Fund (SCF) (UK) July 1994.

37 Although, as we have seen, food consolidation was not designed to force refugees out, one aim was certainly to encourage voluntary repatriation - 'breaking the iron rice bowl' in a much-quoted phrase from the DHC's mission report of 18 June 1993. Zikomo Malawi update No.5 of September 30, 1994 admits that food consolidation may have been misinterpreted.

38 But SCF (UK) alleged that up to mid-July 1994, at any rate, the Malawi radio had not been used to disseminate the message.

39 repatriation from Swaziland, Zambia and Zimbabwe was already well under way.

40 "Thank you Malawi"

41 i.e. Swaziland, Tanzania, Zambia. In South Africa, there seem to have been no assets to dispose of.

42 For example, the area of Nyamithuthu in Nsanje District is almost treeless.

43 Thanks in no small measure to the initiative of the Branch Office, UNDP included a special refugee-hosting area development component in its 1992-96 5th Country Programme with seven sub-components - Roads, Forestry, Health, Water and Sanitation, Education/Vocational Training, Apprentice -based Income Generation and Assistance to Vulnerable Groups. The prospects for funding are, however, understood to be poor.

44 according to the Desk; some UNHCR staff in Mozambique allege it cost as much as half a million dollars and that buses and trucks would have been far cheaper.

45 Repatriation and Reintegration of Mozambican Refugees UNHCR Geneva, May 1993 page 45 para.207 (Rev.2)

46 not to be confused with the South African Catholic Bishops Conference - see below

47 The Appeal also referred to 'a comprehensive assessment mission in the near future'.

48 Estimated at about 5,000 a month during the entire repatriation operation (many of whom returned to South Africa the following day).

49 Operation Hunger had originally been intended to provide the food but withdrew for lack of funds. Fortunately, the South African Government stepped into the breach. (WFP was not represented in South Africa and was not able to begin providing returnee rations until July 1994). UNHCR South Africa had always wanted food distribution to take place in Mozambique, the one issue on which it and UNHCR Maputo were never able to agree.

50 This was the date recommended by the Inter-Agency Needs Assessment mission. Rather strangely, the record of the 13 December meeting says nothing about a starting date.

51 The Committee's record says that the reduction was 'following a survey'. UNHCR Mozambique's printed Progress Report of 27 April 1995 says (para.12) that 'a subsequent study revealed that only some 120,000 persons were of concern to UNHCR'. No copy of the 'survey' has been made available to the evaluation mission which does not doubt, however, that a reduction to some such figure was fully justified. In fact, the Committee's recommendations had been anticipated at an informal Inter-Agency Coordination Meeting on 27 May 1994.

52 A letter from the Chief of the Programme Coordination and Budget Section (PCBS) to the Director of IOM's Department of Operations dated 13 March 1992 agreed to the non-applicability of twelve of UNHCR's standard clauses. When UNHCR Mozambique queried this in October 1994, Headquarters replied that the waiver of standard clauses was still valid.

53 Operational Narrative Report for 1994.

54 And testimony to the operation's overall success. There is usually much less enthusiasm to evaluate failure!

55 which covered the two northern most staging areas.

56 Estimates of refugee numbers were later reduced to 18,000 assisted and 30,000 spontaneous.

57 The May 1993 Plan of Operation said that the luggage entitlement on UNHCR organised transportation would be in a ratio of two buses to one truck (page 13 para.88). This was widely ignored but most easily, one supposes, when railway carriages rather than buses were used for the greater part of the journey home.

58 'Some' in this context means 'approximately' or 'about'. It is meaningless when used to qualify a figure as precise as 14,882!

59 As early as 1991, a UNHCR survey had shown that some 80% of those refugees then in Likuyu came from Niassa Province and the remainder from Cabo Delgado.

60 Both language versions of this and the later information pamphlet have been compared and each is an accurate rendering of the other.

61 Mission report by L. Dakin, Deputy Representative on visit with IOM to Likuyu 20-22 January

62 The Periodic Post Review (PPR) agreed to the creation of 7 posts for Songea on 15 December 1993, subject to the availability of funds. The Co-Chairman complained that many of the requested posts for the Mozambique repatriation had not been budgeted for. The Organization and Methods Section confirms that no earlier request to reopen Songea following its 1992 closure was received.

63 The visit by a 5-man delegation did not, in fact, take place until 13 June but was still prior to registration which began on 30 June.

64 Initially to set up the registration system. This was not adequately done and led to subsequent registration and manifest delays.

65 UNHCR records give a figure of 11,350, TCRS of 11,381, TAMOFA 11,477 and IOM 11,403 (including 3 in December 1994).

66 325 persons actually left Likuyu on 10 September but two (literally) fell by the wayside. One fell out of a truck and both he and another who assisted him were separately repatriated.

67 In contrast, the earlier decision in May that the agricultural package of seeds, tools and fertiliser should be distributed in Tanzania was revised in favour of distribution on arrival in Mozambique.

68 A precedent was perhaps set by a Namibia Repatriation Operation 'Lessons Learnt' Survey in June 1990. Many of the lessons learnt then were relearnt in the context of Mozambique, notably as regards forward planning, transport, communications, information and overall coordination.

69 Problems with CODAN radios in UNHCR Songea vehicles were not finally resolved until 5 October 1994, when the operation was almost at an end.

70 There are other figures (shades of Swaziland) varying from 25,200 down to 21,600. There were also some 800 refugees - perhaps more - spontaneously settled along the border.

71 The refugees are said to have sold 4,000 tons of maize to WFP before their repatriation following WFP agreement to buy their surplus production. This gave them the cash to buy household and other goods to assist their reintegration.

72 But UNHCR Mozambique's Information Bulletin of January 1995 "Mozambique: A Fresh Start" give a different figure of 20,300.

73 By road for the great majority, but with provision (never used) for up to 2,000 vulnerable cases to go by air.

74 Curiously, or by mistake, these figures are the reverse of those given in the May 1993 Plan of Operation - 30,000 by self-repatriation and the remainder by UNHCR.

75 The provisions were a useful guide to other UNHCR Branch Offices which had not yet entered into tripartite agreements or negotiated the 'modus operandi' of repatriation. Particularly useful was the concept of the cross-border travel authorisation document to be carried by UNHCR staff, the staff of implementing partners and drivers.

76 They were lucky enough to take a 6-months ration with them to use up residual camp stocks.

77 In a nutshell, the Branch Office felt that the design and content of the form had been too focused on computer technicalities to meet the requirements of the Information and Communications Systems Section from Headquarters in Geneva, rather than on the individual's repatriation and reintegration needs and on the usefulness of the form in the country of origin as opposed to the country of asylum.

78 Funded from non-UNHCR sources.

79 The MATC team members from 1993 were part of the core group which founded IPA so continuity was assured.

80 The evaluation mission travelled by road on one of the common repatriation routes which runs from Espungabera on the border to Chitobe, a main point of concentration of returnees to Machaza District in Manica Province. Despite 'rehabilitation' by UNHCR/GTZ, it remains in very poor condition. The journey of 105kms took four hours by Land Cruiser, admittedly on a wet day.

81 The camps were never 'closed' and refugees were free to move of their own accord if they chose to do so. Field staff observed that entire families who had been registered for repatriation in the camps had sometimes left by the time they were called to join a convoy.

82 Except occasionally from hunger, when food distribution in their home areas had broken down.

83 Tete Province, the destination of many returnees from Zimbabwe, is not mentioned.

84 a comparison between the potential caseload for the two countries - 120,000 (revised) for South Africa and 100,000 for Zimbabwe - suggests that the allotted time frame for Zimbabwe was disproportionately short and the project target of 20,000 repatriants unduly low.

85 Report by Associate Logistics Officer R.R.W. Muzwidzwa (paragraph 1.2 page 3)