The Journal

The Subjects

The Books





Can we ever rely on refugee statistics?

Oliver Bakewell

At the beginning of 1998 the United Nations High Commissioner for Refugees (UNHCR,www.unhcr.ch) estimated that there were approximately 13 million refugees and asylum seekers in the world, whereas the US Committee for Refugees (USCR, www.refugees.org) estimated that there were 14 million. Although these grand totals are close there are considerable discrepancies in the figures each source offers for any particular country. Taking the example of Gambia, the UNHCR recognises 10,300 refugees, whereas the USCR only counts 8,000. Even the figures from one source may be radically adjusted as new headcounts are undertaken or other information is obtained. Refugee statistics are widely quoted but notoriously inaccurate and in this article I will discuss some of the factors which underlie this prevailing inaccuracy.

Much of what follows is based on personal experience as an aid worker and researcher among refugees in Africa and the illustrations used will show a certain bias to that continent. Since the African continent has the highest number of refugees per head of population and has shared with Europe the dubious privilege of being the setting for some of the largest refugee flows over the last decade, it is an appropriate focus. Although the issue of refugee statistics is very contentious, there has been limited research beyond technical and methodological matters. In a recent paper Crisp (1999) has tried to redress the balance by explicitly looking at the political factors involved in refugee statistics. Here I shall draw on Crisp's paper and also focus on the politics rather than the statistical methods.

Why collect refugee statistics?

Before launching into a discussion of refugee statistics, it is worth pausing and asking why they are important. Put simply, in the current refugee regime, statistics matter as they are a fundamental determinant of the allocation of resources. In any refugee crisis, estimating the number of people involved is one of the first steps in determining the nature and size of any external intervention. Not only are they concerned with the allocation of humanitarian aid but the size of the refugee crisis will also determine the level of political and possibly military resources applied to cope with the situation. The huge numbers of refugees flowing from Kosovo were a very important factor cited to justify NATOs intervention there.

Harrell-Bond et al. (1992) question the assumption that counting refugees is an adequate basis on which to determine need as it is impossible to count refugees accurately. Moreover, they argue that the census of refugees forms 'a central component in an ideology of control' in which the demands of the aid bureaucracy take precedence over refugees' rights. For example, the maintenance of accurate figures on the numbers of refugees is most easily served by restrictions on refugees' movements and their activities. Statistics are gathered by technocratic means in order to enforce an externally imposed notion of fairness and local mechanisms for dealing with shortage are devalued. Using an illustration from Southern Sudan, they argue that 'principles of fairness may direct and inform distribution of scarce resources when people are indeed left to their own cultural devices rather than subjected to an alien and paternalist system' (1992, p. 218).

Statistics are an essential part of the system for accounting for the resources given to refugees. Aid is targeted towards refugees and donors want to know that it has reached the target. This is the case, despite the fact that refugees often settle among poor hosts and it is often very difficult to distinguish refugees from hosts (Chambers, 1986; Hansen, 1979). This is an important factor in the continuing policy of establishing refugee camps even in the face of much research which suggests that refugees and hosts may be better served by more flexible interventions which support local integration (Black, 1998).

A question of definition

Although counting refugees may be a standard task in any refugee crisis, there is still plenty of scope for debate about who should be included within the definition and who is included within any particular statistic. This creates the room for the political manoeuvring discussed below and so the question of definitions is an important starting point.

Throughout history, people have been forced by violence to flee from their homes and the humanitarian principle of offering refuge to them was well established before the 17th century when the term 'refugee' was coined for the Protestants fleeing France to avoid forced conversion. The political upheaval in Europe in the 19th century gave rise to the term 'émigrés' to refer to political revolutionaries who fled persecution. In the 20th century the forced expulsion of people who were then denied their nationality brought in the concept of the 'stateless'. These three terms were largely subsumed within the modern idea of 'refugee' as it has been developed in international legal instruments and the 'international refugee regime' (Zolberg et al., 1989; Wong 1989).

In the aftermath of the Second World War, when millions of people across Europe had been scattered as they fled from persecution, the International Refugee Organization was formed to provide material assistance to refugees and to try to ensure that people were able to return to their countries of origin if desired or to re-establish their connection to a state in some other way. In 1951 it was succeeded by the UNHCR which was mandated to provide international protection to refugees and to seek solutions through voluntary repatriation or 'assimilation' (UNHCR Statute 1950, Article 1). At the same time the UN Convention Relating to the Status of Refugees was drawn up which obliged contracting states to recognise the rights of refugees to asylum, rights of association, freedom of movement, employment and welfare and obliges the contracting state to co-operate with the UNHCR. Within the UN Convention a refugee is defined as a person who,

owing to well-founded fear of being persecuted for reasons of race, religion, nationality, membership of a particular social group or political opinion, is outside the country of his nationality and is unable, or owing to such fear, is unwilling to avail himself of the protection of that country; or who, not having a nationality and being outside the country of his former habitual residence as a result of such events, is unable or, owing to such fear, is unwilling to return to it (Article 1(2)).1

The definition only included those who ran from a particular threat of persecution that was targeted towards them as individuals, and in order to claim refugee status a person had to demonstrate the particular danger they faced. In Africa this proved inadequate, as there were mass movements of people fleeing from the general upheaval of war as independence struggles flared up during the 1960s. The Organization of African Unity extended the provisions of the UN Convention to those who were forced to flee owing to external aggression, occupation, foreign domination or events seriously disturbing the public order (OAU Convention Governing the Specific Aspects of Refugee Problems in Africa 1969, Article 1(2)). A similar regional definition was made for Latin America in the Cartagena Declaration of 1984. Thus, a person recognised as a refugee in Africa would not necessarily be accepted as a refugee in Europe.

These formal legal definitions should be distinguished from those in popular use, particularly in the media, which tend to denote anybody who has had to flee from home as a refugee, even if they stay within their national borders.2 The distinctions become less clear when the borders change as dramatically as they have in Europe and the former Soviet Union over the last decade. Usually there is some association with violence but there is a growing trend to use terms such as 'environmental refugee' to describe those forced to leave because of changes in circumstances beyond their control.

The politics of refugee statistics

Putting aside the definitional problems, in any particular situation where there is general agreement about who should be considered a refugee, there is unlikely to be any consensus on the numbers of people who fall into the agreed definition. As noted above, refugee statistics are required for a variety of reasons in different contexts, but in every case they are used to make decisions about the allocation of resources. They are, therefore, an arena of struggle and reflect the interests of parties involved. In this section I will briefly discuss some of the interests and how they may affect refugee numbers in different ways.


Refugees have a special status within international law and with it go rights and privileges in terms of permission to stay and humanitarian aid. Programmes to assist refugees, in both industrialised and developing states, tend to put a lot of effort into restricting access to these benefits in order to weed out those who are not entitled. There is a widespread perception that people will look for ways to get round the system and inflate the refugee numbers.

In developing countries or collapsed states (such as the former Yugoslavia), when there are mass flows of refugees, humanitarian aid may be delivered in the form of food, water, shelter material (usually plastic sheeting), kitchen equipment and other such resources. At times these will be given in volumes which might be significant compared to the resources available to local people who are not refugees. If refugees cross the border at a manageable rate, they will be registered on arrival and given a ration card, usually covering a family rather than individuals. This ration card will have to be presented at every distribution. Clearly at the point of registration refugees have an incentive to inflate the size of their household or if possible to register twice, a practice known as 'recycling'. Inevitably a market in ration cards will develop, particularly if refugees move out of the camps to stay elsewhere (often illegally) or decide to repatriate. Some local people may also present themselves as refugees, especially where they share a common ethnicity with their neighbours from across the border. This is frequently the case in Africa, where national boundaries often cut across ethnic groups and are weakly controlled (Bakewell, 1999). Figures produced from the initial registration will become outdated without some system of recording births, deaths and movements out of the area. In an emergency such systems are unlikely to be a priority and there is little incentive for refugees to co-operate, especially to report deaths and migration which will decrease the household's entitlements.

Within camps or other segregated settlements the accuracy of the registration figures can be monitored by such means as counting the number of shelters, and when the recorded and actual numbers of refugees appear to be to far apart, a new mass registration may be required. If refugees are not registered on arrival, then a registration exercise will be carried out once the situation has stabilized sufficiently to allow it. Mass registration is likely to be a highly contentious issue as it determines the distribution of resources, and with that power, within the camp. It can be seen as an attempt by the donor community to seize back control.

The case of the Goma refugee camps for Rwandan refugees in 1994 clearly illustrated this. When refugees went over the border into Zaire in July 1994, it was impossible to register them and initially food was sent into the camps and community leaders controlled its distribution. Many of these leaders were seen as people connected with the genocidal Government and military and it was very clear that some of the most vulnerable people were not receiving food and some was being diverted to the military. There were only estimates of the numbers of people in the camps and the volumes of aid were based on these. However, it was widely recognised that these estimates were inaccurate and considerably overstated the number of refugees. Donors and aid agencies perceived registration as a mechanism for efficiently using resources and ensuring all refugees received rations. Refugee leaders, however, resisted the count and it was postponed repeatedly until they were persuaded to co-operate in January 1995.

Such scenarios suggest that refugees have an interest in declaring themselves in exaggerated numbers and inform the discourse of refugees as clients under the patronage of the international aid regime, in which refugees are to be controlled and systems have to be devised to prevent 'cheating'. However, not all refugees seek humanitarian aid; many prefer to stay outside its reach. In Africa the majority of refugees remain outside refugee camps or settlements and receive no humanitarian aid (Kibreab, 1991). Those who can make their own way do so and many do not want to come to the attention of the authorities, especially if it may result in their having to move to camps. In remote rural areas far from the reach of the state, it may be possible for large numbers of refugees to blend in and settle among local hosts (Hansen, 1979).

In industrialised states, with much stronger border controls and individual assessment of asylum applications, the key resource available is refugee status itself, which effectively grants the refugees the right to remain indefinitely. This has become more valuable because other avenues for immigration into industrialised nations from developing countries have been steadily closed down over the last twenty years since the economic downturn in the 1970s and the decreasing need for migrant labour. As a result, potential migrants have a stronger incentive to claim asylum as the only channel offering them a chance of access.

States of origin

No state is likely to welcome a haemorrhage of its population across its borders fleeing violence. At best it shows that the government is failing to provide the protection and security which citizens expect. At worst, when the state is perpetrating the violence, the flight of refugees will expose its actions to the world. In either case, the government will have no means of counting how many people leave and any claims to know the number of refugees will be treated with scepticism.

In their efforts to deny the flow of refugees, states may question the motivations of those leaving, by strategies such as associating them with rebel movements who pose a threat of military invasion. The refugee is portrayed as the enemy. At times, such claims have an element of truth and some refugees have contributed to military forces gathering in neighbouring countries (such as Afghan refugees in Pakistan contributing to the Taleban movement and Rwandan refugees in Zaire harbouring a government in exile). However, such 'refugee warriors' (Zolberg et al., 1989) form a minority in refugee populations whose main concern is survival rather than politics.

Another way of playing down refugee movements is for countries of origin to deny their existence. This has been seen most blatantly in 1999 when the world's media photographed and interviewed refugees from Kosovo in Albania while Serbian officials denied their existence and claimed the people in the media view were acting. An alternative method of denying refugees' existence is to claim that the people leaving were never really citizens in the first place (Crisp, 1999).

Host states

The majority of refugees are found in the poorest countries of the world with very low average incomes, minimal healthcare and education, and low life expectancy. The arrival of refugees without any resources and in need of immediate assistance may create an impossible burden for such states, which struggle to meet the needs of their own people. Faced with such conditions, poor host states may appeal to the international community for assistance. However, there is no standard formula for determining what help may be forthcoming. The response of donor states will depend on their relationship with the refugees' country of origin and the host state and also the circumstances of the refugees' flight (for example, whether it has come to the attention of the public through media coverage). Whatever assistance is forthcoming will be scaled according to the number of refugees and there is a widespread expectation that host states will inflate the figures to maximise the resource flow.

Humanitarian aid for refugees will not only provide a range of materials for distribution but they may also generate considerable investments in jobs, contracts and infrastructure. The host government's degree of control over these resources may vary considerably depending on its policy and capacity, but there is always room to exert some leverage over the use of aid to refugees and increased numbers will raise the amount available. As Crisp (1999) points out, frequent over-reporting by local government officials may be a response to the chaotic situation of an emergency. Faced with a large crowd of refugees who have immense needs it is always safer to err towards the more generous estimates of their numbers to increase the chance of sufficient aid arriving. Given that very often supplies move slowly and in smaller quantities than originally planned, using generous planning figures can reduce the suffering of the refugees.

Besides such direct concerns about resources, host governments may have other interests which affect their reporting of refugee statistics. If the country of origin of the refugees is seen as a friend, its neighbours may play down the number of refugees in order to avoid embarrassment to it. If the country of origin could pose a potential threat to the hosts, the latter may reduce the number of refugees to avoid retaliation (Kibreab, 1991). For example, frontline states under-reported the number of South African refugees in order to lessen the chances of military reprisals. In a related case, one factor which delayed Malawi's acknowledgement of Mozambican refugees fleeing from Renamo was its wish to avoid upsetting Renamo's sponsors, South Africa (Zetter, 1995). In contrast, where the host country is hostile to the regime in the country of origin, the numbers of refugees are likely to be inflated to bolster the case against it.

In industrialised states where refugees come in small numbers and the states are wealthy enough to meet the needs without external aid, the numbers game is more likely to be played to a domestic audience. During the Cold War refugees from communist states arriving in the West offered welcomed propaganda opportunities. Today, rather than deploying propaganda against a rival ideology, it is used against the asylum procedure itself. As countries look to restrict access to asylum, there is a tendency to highlight the numbers of refugees and asylum seekers as a justification for ever more stringent measures. Distinctions are frequently blurred between asylum seekers, economic migrants and illegal immigrants and the focus is often on the numbers of asylum applications rather than the small numbers who are actually granted refugee status.3

The UNHCR and other humanitarian agencies

The UNHCR has no statutory funding and is reliant on voluntary contributions from donor nations. Frequently the cash is not forthcoming and proposed budgets are slashed as hoped-for donations do not materialise or funds (and other resources) are more thinly spread to cope with new emergencies. Likewise, other humanitarian agencies, mostly non-governmental organizations (NGOs) working with refugees are dependent on funding either from public donations, grants from governments or other grant-making bodies. In all cases, the agencies involved have to make a case for receiving the funds and then demonstrate they have used them properly. A starting point for any programme is to know how many refugees are involved.

Clearly such statistics are not only necessary for funding and accounting purposes but also important for operational reasons. Gathering such information is a high priority in any assessment mission and the data will inform the programme strategy. However, obtaining accurate numbers is very difficult and, in an emergency, there are likely to be competing figures bandied about by all parties. For aid agencies there must be an inclination to err on the side of generosity in the initial assessment in order to increase their chances of meeting refugees' needs, but it is not acceptable practice to deliberately inflate the figures to boost funds. They are more likely to focus attention on the vulnerability of the refugees and to encourage an emotional response from donors (Harrell-Bond et al., 1992).

The UNHCR is in a more difficult position than NGOs as it is subject to more explicit political pressure in determining its interventions. To a large extent, the UNHCR has to agree on the number of refugees with the host country and country of origin and placing contrary figures in the public domain can cause offence to the countries involved, with implications for the UNHCR's operations there. This was particularly evident in the case of the massive forced repatriation of Rwandan refugees from Zaire in 1996, during which some 400,000 refugees went missing in the bush where they were being killed by Zairean rebels in league with Rwanda. When Oxfam and the UNHCR raised the issue, the Rwandan Government angrily denied the refugees' existence and relations between the UNHCR and Kigali were extremely stretched (Pottier, 1999). At other times, when the UNHCR has played the governments' games, it has been caught out when the official planning figures become self-evidently wide of the mark (Harrell-Bond et al., 1992).

Since it is the refugee crisis which brings in the funds and some organizations, particularly the UNHCR, have a mandate limited to working with refugees, agencies have to target their programmes at refugees. In Zambia, for example, where the majority of Angolan refugees live outside formal settlements and many have effectively integrated into Zambian society, the UNHCR focuses nearly all of its resources on settlements where refugees can be easily identified and assisted. Any activity outside the settlements has to be justified by reference to the number of refugees in the general population in order to designate an area as refugee-affected and an appropriate area for work. The main contact that self-settled refugees have had with the UNHCR over the last five years has been various attempts at counting them, all of which have failed. These registrations have tried to make a distinction between refugees and hosts which has largely dissolved making the task impossible. Worse still it is counter-productive as it may undermine the local integration of refugees (Bakewell, 1999). In such situations, registration becomes an exercise to generate numbers for donors rather than giving useful information for planning assistance to refugees.


Refugee statistics may justify the contribution or intervention by donor governments but they are moulded by their particular interests. The numbers can be crunched in appropriate ways to tell a particular story using similar tactics to those described for countries of origin or hosts. For example, during the Cold War western states were happy to be hosts to refugees from communist countries and they were also willing to act as donors to support refugees from communist states such as Afghanistan, Ethiopia and Vietnam. Such assistance could also act as cover for support of opposition groups in exile (Crisp, 1999).

More generally donors may have interest in minimising the numbers of refugees in order to reduce their obligations, especially after the initial crisis is deemed to have passed. After the Dayton Peace Accord was signed there was pressure from donors to reduce quantities of food aid for Bosnia by cutting the numbers of beneficiaries through more precise targeting (Crisp, 1999). In the case of the former Yugoslavia, where donors have been involved in finding political (or military) solutions to the refugee crises, they will also have an interest to see the numbers of refugees reduced sharply as an indicator of the success of their policies.

The practicalities of refugee statistics

These political factors are particularly important in determining which refugee statistics are given credence as there is no simple and effective way of counting them. Even industrialised states have great difficulty in clearly stating how many refugees are living in the country despite the resources and sophisticated techniques that they have at their disposal for gathering the information. Not surprisingly, it is harder to get reliable data in developing countries. In the chaos of an emergency, estimates of numbers may come from many sources as people enter across borders into many different areas. Access to areas may be difficult because of their remoteness, security concerns or government obstruction. People may be counted twice or not at all and rough estimates will be all that is possible.

Even once things have settled down and registration is possible, things do not necessarily get more simple. As shown above, there are so many interests involved in counting refugees that there is no reliable methodology for achieving an accurate count. The necessity of counting is taken for granted by external agencies, and very often they will assume that the refugees will be out to 'cheat' the system. Therefore, they are sceptical of any self-reported count and demand more 'scientific' methods (Harrell-Bond et al., 1992). These methods sometimes appear more appropriate for counting animals rather than people who have a right to dignity. Large numbers of agency staff (perhaps excluding local staff for fear of corruption and using only expatriates) may sweep through a camp within as short a time as possible 'branding' them with some mark (for example, gentian violet, an antiseptic ointment, is often used). Refugees may be ordered to stay in their shelters and, in order to catch them before they move during the day, registration may start at dawn. Those who are away from the camp at the time of the registration may find themselves without ration cards and hence without food. Outside of camps enumerating refugees is even more difficult, especially if they are not allowed to stay outside. Hansen (1979) describes how self-settled Angolan refugees in Zambia fled into the bush whenever they heard a government vehicle for fear that they were coming to register them. Not surprisingly, given their exclusion from the process and the degrading methods used, such registrations do nothing to encourage refugees' co-operation.


Given that there are no reliable ways to enumerate refugees and that there are so many interests involved, it is important not to have high expectations of the quality of figures that are put in the public domain. The numbers always need to be put in the context of where they have come from, how they have been generated, who is using them, and who is counted as a refugee. Because so many of the figures are presented through the media, with its reputation for exaggeration, it may be tempting to assume that the figures bandied about are overstating the case. However, there are many interests which may tend to play down the figures and general rules are dangerous. Refugees themselves are largely excluded from the counting process except as its objects. They are exposed to the much more dangerous situation of vital resources being allocated to them according to a highly subjective, political system over which they have no control.


  1. A slightly different definition is used in the UNHCR Statute, which also refers to a refugee being a person unable to return for reasons 'other than their personal convenience'. The original instruments, both the 1950 Statute and 1951 Convention, restricted the definition to those forced to move by events prior to 1st January 1951. The Convention was amended by the 1967 Protocol, which removed this date restriction.
  2. Such people are widely referred to as internally displaced people (IDP) and, given that they exist in the same conditions as refugees except for the fact they have not crossed a border, there is a growing debate about their exclusion from the legal protection and aid of the international community (Hampton, 1998).
  3. Between 1991 and 1995 there were 2.4 million asylum applications in Europe of which 11% were granted full refugee status and a similar number given leave to stay on other grounds (UNHCR, 1998). During the 1990s in Europe there has been a growing tendency to offer the weaker status of temporary protection to those seeking asylum from the former Yugoslavia.


Bakewell, O. (1999), Refugees Repatriating or Migrating Villagers? A Study of Movement from North West Zambia to Angola, Unpublished PhD, University of Bath.

Black, R. (1998), 'Putting Refugees in Camps', Forced Migration Review, 2, pp. 4-7.

Chambers, R. (1986), 'Hidden Losers? The Impact of Rural Refugees and Refugee Programs on Poorer Hosts', International Migration Review, 20(2), pp. 245-263.

Crisp, J. 1999, 'Who has counted the refugees?' UNHCR and the politics of numbers, New Issues in Refugee Research, Working Paper No. 12, UNHCR, Geneva (www.unchr.ch)

Hampton, J. (ed.), (1998), Internally Displaced People: a global survey, Earthscan: London.

Hansen, A. (1979), 'Once the Running Stops: Assimilation of Angolan refugees into Zambian border villages', Disasters, 3(4), pp. 369-374.

Harrell-Bond, B. Voutira, E. and Leopold, M. (1992), 'Counting the Refugees: gifts, givers, patrons and clients', Journal of Refugee Studies, 5(3/4), pp. 205-225.

Kibreab, G. (1991), The State of the Art Review of Refugee Studies in Africa, Uppsala Papers in Economic History, Research Report No. 26.

Pottier, J. (1999), 'The "Self" in Self-Repatriation: Closing Down Mugunga Camp, Eastern Zaire', in Black, R. and Koser, K. (eds), The End of the Refugee Cycle? Refugee Repatriation and Reconstruction, Berghahn Books: London.

Wong, D. (1989), 'The Semantics of Migration', Sojourn, 4(2), pp. 275-285.

Zetter, R. (1995), 'Incorporation and Exclusion: the life cycle of Malawi's refugee assistance programme', World Development, 23(10), pp. 1653-1667.

Zolberg, A.R. Suhrke, A. and Aguayo, S. (1989), Escape from Violence: Conflict and the Refugee Crisis in the Developing World, Oxford University Press: NY.

Oliver Bakewell
27 Tynemouth Road
London N15 4AT

E- mail:oliver@bakewell.fsnet.co.uk

Journal 069 Index Top of page