Cover Image
close this bookMedical Assistance to Self-settled Refugees (Tropical Institute Antwerp, 1998)
close this folder4. Food aid
View the document(introduction...)
View the documentA rational basis for deciding on food aid?
View the documentRefugee registration
Open this folder and view contentsAssessing the nutritional situation
Open this folder and view contentsDelivering food aid
Open this folder and view contentsOf evidence and pressure


Neither food output, nor prices, nor any other variable like that can be taken to be an invariable clue to famine anticipation, and once again there is no substitute for doing a serious economic analysis of the entitlements of all the vulnerable groups.

Amartya Sen

In unstable situations, morbidity and mortality often increase considerably.2-6 Whether and to what extent food shortage, malnutrition and increased mortality are causally linked remains a matter of debate,7-12 but the importance of food goes far beyond physical survival, even in emergencies. Although during the emergency phase water and shelter may be more important for short-time survival,13 food is indeed one of the priorities. Food shortage and access to food are invariably important issues in unstable situations.9,14,15 Aid agencies often consider food aid as the most urgent need of refugees.

Food aid in times of food scarcity intends to reach those without the means to obtain it. Food aid does not, however, take place in a vacuum, but in a social context, even if society is in disarray.16,17 Food is one of the main economic assets in countries in crisis. It is not because food aid is supplied free ('not to be sold or exchanged'), that it loses its economic value. The poorer and the more disrupted the society, the higher the relative value of food. Not surprisingly, some may try to misappropriate food aid, not only to feed themselves or their community, but also to gain economic benefit. Food aid also influences local offer-and-demand balances. Food scarcity raises food prices, to the benefit of those offering food on the markets, mainly merchants, but sometimes producers too. These are potential losers if food aid effectively reaches those in need.

During armed conflicts, poorly supplied fighting forces may claim a share of the food intended for civilians. Food aid distributed to refugees may be transported across borders to supply guerrillas, or the civilian populations that remained behind. Food may even be used as a weapon of war.18 Armies may try to starve certain areas for military purposes, destroy or loot food stocks and crops, and hinder food aid.

Within such contexts, all aspects of food aid are highly problematic.19 First, the quantity of food aid is always a balancing act between supplying enough to feed those in need, and limiting supplies so as not to disrupt the local markets. Second, if it is still relatively easy to decide when food aid should start, or when it is insufficient, it is considerably more difficult to know when food aid should decrease or stop altogether.20 Third, the identification of beneficiaries is another major problem. Registration is often incomplete and fraught with difficulties. Distinguishing those in need of food aid from those not in need is always somehow arbitrary. Distinctions are often based on value judgements that may not be the same for those delivering and those receiving assistance.21,22 Fourth, logistic constraints often hamper transport and distribution. Lastly, appropriate distribution channels are difficult to design and control.

Because of these difficulties, food aid is more often than not contentious,23 it is the most visible benefit granted to refugees, and conflicts may erupt or crystallise around it. It is a source of conflicts between humanitarian agencies and refugees, among humanitarian agencies and among refugees, between refugees and hosts, between humanitarian agencies and merchants, between humanitarian agencies and donors, between donors and host governments.17,24,25

Humanitarian agencies have gained considerable experience with food aid to refugees. Manuals describe how to organise refugee registration, anthropometric surveys, food distribution, food basket monitoring, as well as supplementary and therapeutic feeding programmes.19,26,27 This know-how is, however, largely based on experience within relatively closed systems, such as refugee camps, where beneficiaries are entirely dependent on food aid. Strategies used are not necessarily appropriate for open refugee situations, where beneficiaries of food aid (the refugees) live integrated, both spatially and economically, with non-beneficiaries (the hosts). This makes selection and registration of beneficiaries considerably more difficult. Some refugees may be self-sufficient, while some hosts may not. In open situations, nutritional programmes invariably induce system effects in the wider society, and affect local balances.28

Throughout PARLS, food aid was contentious and fraught with difficulties. At no point in time, satisfactory solutions were found. This chapter analyses decision making in food aid, and reviews the experience in Guinea with refugee registration, with assessment of the nutritional situation, and with food aid.

A rational basis for deciding on food aid?

THE DECISION-MAKING PROCESS. Figure 16 represents the essential ingredients and the logical sequence of what would be a rational basis for decision making in food aid. Assessing the number of refugees usually requires refugee registration through an exhaustive census, but is often based on guesstimates. Humanitarian agencies usually carry out anthropometric surveys to know the nutritional status of the refugees. Information on the number of refugees and their nutritional status, together with expertise in food aid, yields an estimate of the food needs of the refugees. Based on this knowledge decisions on food aid are taken (quantity and quality of food needed, modes of purchase, transport and storage, distribution channels and control methods). When implemented, these decisions result in distribution of food aid to refugees to improve or maintain their nutritional status. Periodic anthropometric surveys and checks on refugee numbers allow for monitoring feedback of the situation, and adjustment of decisions.

Figure 16: Decision making in food aid for refugees

CLOSED VS. OPEN SYSTEMS. For food aid to refugees, or to 'beneficiaries' more in general, there are two possible situations. The beneficiaries may be considered living in a closed system, cut off from the rest of society, entirely dependent on food aid and in need of full rations. This is the case in isolated refugee camps, where refugees have no access to land or labour opportunities. Alternatively, the beneficiaries may be considered living in an open system, partially self-sufficient, and thus only partially dependent on food aid.

In the situation of complete dependence, decisions on the quantity of food aid needed are relatively straightforward. All beneficiaries need a 'full radon' of 2,100 kcal per person per day, with a minimum of 10% of calories supplied by protein, and the essential micronutrients. The need for food aid does not decrease with time, and does not depend on the season or on good or bad harvests. The only important variable is the number of refugees, which may change with new arrivals or departures. Refugee registration has thus to be updated regularly. If food aid is delayed or insufficient, in quantity or in quality, nutritional problems, particularly micronutrient deficiencies,29 soon appear.30,31

When refugees are only partially dependent on food aid, as was the case in Guinea, decisions are more difficult. Insights in the availability of food in the society at large, and in the degree of self-reliance of the refugees, are needed to estimate the quantity of food aid needed, and the groups in need. The general refugee population and those identified as being vulnerable, often receive different rations. Coping mechanisms and needs of refugees evolve over time. Knowledge on local food production and local markets should play a role in decisions on food aid. If, however, the food distributed lacks essential micronutrients, this rarely causes problems. If food aid is delayed, this does not immediately lead to widespread malnutrition, as refugees have other coping mechanisms.

DEVELOPMENT ACTORS & RELIEF ACTORS. In comparison with most other refugee situations the number of different actors involved in the relief system in Guinea was fairly limited. During the first months of the refugee influx, the role of central government, UNHCR and WFP was minimal. UNHCR and WFP, who have an international mandate in refugee situations (Box 2), progressively played a more prominent role.

Medical aid to the refugees was mainly developed and carried out by the regional office of MOH and those agencies that were already assisting MOH before the refugee crisis: MSF, Mission Philafricaine and GTZ. The situation as regards food aid was more complex. Specialised agencies started operations in the Forest Region to implement food aid for the refugees. UNHCR and WFP took the lead. Also Red Cross,* the refugee committees, and several NGOs intervened at various steps in the process (Figure 17). UNHCR, Red Cross and the government were involved in the registration of the refugees. MOH and the medical agencies assessed the nutritional situation. With this information, and using their expertise, UNHCR, WFP and the government estimated the food needs. These same actors, together with the donors, decided on food aid. The different implementing agencies influenced decisions on practical modalities in their respective fields. WFP imported and transported food aid to regional stores. Adventist Development and Relief Agency (ADRA), and later GTZ, managed regional stores and transported food to local stores and distribution points. The Red Cross, in collaboration with local refugee committees, distributed the general food rations to the refugees. Eglise Protestante Evangque and Jesuit Refugee Service targeted unregistered and vulnerable refugees. MOH and the medical agencies organised feeding programmes for malnourished children. Logistic constraints and lack of commitment of donors often prevented that decisions were effectively materialised.

[* The Guinean Red Cross was not active in the Forest Region before the arrival of the refugees. In 1990, it recruited volunteers, and started operations assisted by the Federation of Red Cross Societies. In the text, 'Red Cross' means 'Guinean Red Cross, assisted by the Federation'.

Throughout the refugee-affected areas, UNHCR set up elected refugee committees as representatives for the refugees. But the election process was not clear to most refugees, and many thought that the members of the refugee committees had been appointed by UNHCR and were receiving benefits from UNHCR for sitting on the committee. Many refugees felt that the refugee committees were part of an alien bureaucracy ('the refugee system') and did not serve the interests of the refugees.53]

Box 2: UNHCR's central position in the refugee system

After the first World War, with the construction of strong nation states in East Europe and the large flow of refugees this caused, an international refugee regime was created in Europe.32 Since then, international conventions have consolidated this regime, and UNHCR has been created. It received an international mandate to protect and assist refugees worldwide.33 UNHCR works in close collaboration with host governments, who remain sovereign on their own territory. For food aid, UNHCR collaborates with WFP. WFP is responsible for purchasing and transporting food, while UNHCR is responsible for food distribution and assessment of the needs. UNHCR and WFP usually establish contracts with implementing partners. In refugee situations in sub-Saharan Africa, international humanitarian organisations (e.g. Oxfam, MSF, Caritas, Red Cross associations, &c) and NGOs of the host country play an increasingly important role, at the expense of national governments. In some refugee situations, there are tens of NGOs, many of them active in a limited sector (e.g. health, education, care for unaccompanied minors, &c.). UNHCR and the host government co-ordinate activities, but each NGO has somehow its own motives, agenda, institutional logic and timeframe.

MOH and its field partners were already operational in the Forest Region before the arrival of the refugees, but had to expand considerably their staff and logistic means for PARLS. All other agencies came to Guinea, especially to assist the refugees. This influenced their attitude, judgements, priorities and time-perspective. The agencies already present had development of the Forest Region as primary objective. They can be called 'development actors'. These agencies perceived the refugees first as an unplanned interference with their development activities. For the other agencies, the 'relief actors', refugee relief was their prime objective. For them, relief aid had, somehow, to be inserted in the host society. Even within agencies, especially MSF, these two types of logic co-existed, and sometimes conflicted

Figure 17: Actors in food aid in Guinea

CO-ORDINATION. UNHCR established an office in the capital, Conakry, to co-ordinate its activities in Guinea, and to consult the central government. In Conakry, a national refugee co-ordination committee was established, representing different ministries and law enforcement agencies. This committee decided to limit refugee assistance to the Forest Region. Beyond that, its role remained limited. The regional government and UNHCR assumed overall responsibility for PARLS in the Forest Region. UNHCR established sub-offices in N'Zkornd Guu, and held regular co-ordination meetings with relevant government bodies and implementing partners. These meetings served mainly to inform the Guinean authorities on developments within PARLS. In each prefecture, the prefect was theoretically responsible for PARLS. However, in practice, the foreign agencies often called the shots.

Refugee registration

Although this has been challenged21 refugees are commonly counted registered, both for identification and protection, and for material assistance. For humanitarian agencies, counting the refugees and establishing the 'refugee caseload' is central in planning and fundraising, and for their accountability.34 UNHCR has a mandate to protect and assist refugees, and systematically distinguishes between refugees and hosts. Also in Guinea, registration and ration cards were central to the refugee system. However, the fact that most refugees self-settled among the host population, spread over a large area and were very mobile, both within and across state borders, complicated registration considerably. The registration process evolved over time (Figure 18).

LOCAL COUNCILS. In early 1990, local councils counted the number of refugees arriving in their administrative territory. During the first months, there was probably no deliberate exaggeration of actual refugee numbers, only a lot of confusion. When refugees passed through one village and settled in the next, they were often registered twice. When the local councils forwarded new data, it was often unclear whether this concerned new arrivals or updated totals, whether it was the total number of refugees, or the number of households. Later, when UNHCR announced that quantity of food aid would depend on the number of refugees, numbers were inflated, sometimes openly increased during meetings. At that time, the UNHCR official leading the assessment mission tolerated, if not encouraged this. Early 1990, the refugee-influx in Guinea was occurring without media attention and without interest of the international community. UNHCR had to find arguments to make a case for the Guinean refugee crisis. In the absence of any dramatic emergency, only numbers could convince.

Figure 18: Refugee registration: official data vs. guesstimates, 1990-95

RED CROSS. In April-May 1990, UNHCR needed a more reliable refugee registration for distribution of food aid, and asked the Red Cross to register refugees, and deliver ration cards. In the Red Cross registration system, refugees had to constitute groups of 50, which were then registered as a group. Each head of household within the group of 50 received one ration card. These ration cards - commonly called rice tickets - served mainly to verify the authenticity of refugees' identity during food distributions, and to obtain free medical care.

PROJET DE RECENSEMENT. In 1991, UNHCR and the Ministry of Planning established the Projet de Recensement to improve refugee registration. Despite the recognition of structural problems in the Red Cross system, the Projet de Recensement largely adopted the Red Cross data and continued to implement their registration system with little modification.

FALSE POSITIVES. Since the start of food distribution, refugee registration had entitled all refugees to food rations, without any assessment of need.

Part of the problem with registration was that there was only one system, combining identification, registration and entitlement to benefits in one process and on one card. As registration was mainly a means to get access to food aid, there were incentives to register multiple times. This has also been observed elsewhere.17,34 In 1991, a report highlighted major problems with the registration system. There were duplicate registration numbers, altered cards, cards with no names and forged cards.35* Heads of families registered multiple times, married women registered as heads of households and English-speaking Guineans registered as refugees. The registration teams were also under pressure from local influential Guineans to register Guinean families. Refugees deciding to return to their country, often left their ration card with their hosts or fellow refugees, or sold it. At certain moments, there was a real market in ration cards. Such individual abuses were, however, insignificant compared with the massive fraud committed by merchants and government officials. Some owned dozens of ration cards for 50 beneficiaries each. It was rumoured that they were registered for FG5,000 (US$5) per card.36

[* UNHCR found many cards with identical lists, figuring the 50 same names, others were composed of fictive names (e.g. 50 jazz musicians).]

FALSE NEGATIVES. Despite these false positives some bona fide refugees were not registered. In 1995, one NGO conducted a house-to-house survey and found that about 15 per cent of the total refugee population was unregistered and did not receive food assistance.37 Some refugees claimed that officials had removed them from assistance lists without explanation or had refused to register them, unless they paid a bribe.36 Officials asserted that such claims were a method refugees used to register multiple times. Registration of new arrivals was also problematic. To reduce the possibility for families to register multiple times, refugee registration was often stopped and periodically resumed to register large influxes of new refugees. Small groups of new arrivals could not get registered. Later, UNHCR even decided not to register new arrivals unless they agreed to relocate in refugee camps (Box 1).

RE-REGISTRATIONS. From late 1993 on, UNHCR decided to proceed to re-registrations. Registration officers went on very short notice to refugee settlements, called all heads of households and verified registration. After each re-registration, however, there were many complaints from genuine refugees that they had been barred on arbitrary criteria. Most notably, refugees registered in certain settlements, but not physically present during re-registration, were barred from the lists. The result of these re-registrations was a decrease in the overall number of beneficiaries, mainly in the cities. However, many false refugees continued to have ration cards, as they bribed the registration officers to remain registered.

[ This verification was done in different ways. Sometimes the head of household was requested to name all dependants, and this was checked-off against the initial registration data. At other occasions, the groups of 50 beneficiaries had to be present physically.]

OFFICIAL REFUGEE DATA VS. GUESSTIMATES. The official refugee data suffered from these shortcomings in refugee registration. During 1991 official numbers of refugees in N'Zkority increased considerably, although no new refugees had arrived. Between 1992 and 1995, the official number of refugees remained stable around 570,000 - 580,000 (Figure 18), although 150,000 new refugees arrived during this period, and there were only few official repatriations. During 1996, considerable numbers of refugees returned to Liberia. Official refugee data did not adequately reflect this either. Figure 18 compares official refugee data with guesstimates, based on observed refugee influxes and discrepancies between registration data and survey data. During 1991-93, it was estimated that up to one-third of all registrations were false, in the cities this was close to 50%.

FOOD FRAUD. The false registrations resulted in massive misappropriation of food aid by Guinean merchants.* This food was sold on the markets and depreciated the rice price. Misappropriation of food aid discredited PARLS in the eyes of donors, relief agencies, and even the refugees. This bad reputation was probably the main cause for the reductions in food aid imposed during 1994-95, which resulted in real famine in certain rural areas in 1995. By 1995, re-registration and a stricter registration system had corrected much of the false registrations, but by then PARLS had acquired its bad reputation. This bad reputation probably created the mind-set for UNHCR officials to impose new arrivals to settle in camps, so as to separate them from the old refugees.

[* Some Guineans also used refugee rations cards to obtain free medical care in health centres and hospitals, where they had to pay. However, reports on such abuses remained anecdotal. Health care can indeed only be consumed by one individual at a time. Moreover, in the process one gets registered and is observed by health workers and other patients, who are not part of the refugee system, and often disapprove abuses. Anyhow, no actor in the refugee system ever considered that misuse of ration cards to obtain free medical care constituted a serious problem.]


Next to counting and registering refugees, the other classical ingredient of decision making in food aid is the assessment of the nutritional situation. This is generally considered a priority in unstable situations.38,39 In refugee camps, humanitarian agencies routinely perform standardised anthropometric surveys to assess the prevalence of malnutrition among the refugees.26 In more open refugee situations anthropometric surveys are not considered sufficient, and availability of food should also be evaluated to get insight in the overall nutritional situation.19 Assessing the nutritional situation is supposed to help decision-makers to estimate food needs, and thus make appropriate decisions on how much food should be provided, to whom and through what channels.

Figure 19: Presumed relations between nutritional assessment and food aid in PARLS

ASSUMPTIONS. Figure 19 shows how the information gathered was to inform decision making. The underlying assumptions were: (1) that anthropometric methods would produce valid indicators of the extent and trend of malnutrition in the community, and (2) that together with the data from the refugee registration, and insight in the local food economy ('expertise in food aid') this would lead to correct estimation of food needs and good decisions on food aid for the refugees. These assumptions were rarely made explicit, and their validity can be questioned.

THE SEARCH FOR APPROPRIATE APPROACHES. The approaches used evolved over time (Figure 20). Throughout 1990-96, MOH, MSF and the other actors have been trying to find the best approach for measuring the extent and trend of malnutrition, and the availability of food. In the beginning of PARLS, market prices of food commodities were monitored as an indicator of the evolution of food availability in the community. The collection and interpretation of data met with serious difficulties, and was abandoned after a few months.

In terms of anthropometry, three approaches were used to gather relevant information: (1) large cross-sectional anthropometric surveys, (2) intake anthropometric surveys of new refugees, and (3) monitoring of the proportion of malnourished children at the curative clinic (PMC monitoring) followed by small targeted anthropometric surveys. Table 11 summarises the anthropometric methods used. At no point in time, the 'relief actors', be they UNHCR, WFP or NGO field staff, nor the 'development actors' were fully satisfied with the approach used.

Figure 20: Methods used to assess the nutritional situation, 1990-96

The use of different anthropometric methods and indicators over the years illustrates the continuous search for the best approach to nutritional assessment. It is also an illustration of the conceptual and practical difficulties that arise when 'development actors' and 'relief actors' work together, and have to find compromises between their different types of logic. The nutritional information gathered through considerable efforts rarely supported decision making, and mainly served as a post hoc justification of decisions already made. The following section will describe and discuss the different assessment methods used and the results this yielded.

Table 11: Anthropometric methods used for nutritional assessment in PARLS, 1990-96



Criterion for malnutrition in a child*

Indicator for prevalence of malnutrition

Assess the extent of malnutrition in the community

Anthropometric surveys of children between 6 and 59 months;b cluster sampling or exhaustive surveys

Four different criteria have been used:
<80% median W/H, or oedema
<-2 z-score median W/H, or oedema
MUAC <12.5 cm, or oedema
2-stage MUAC-W/H: first, MUAC for all children, followed by W/H for those <13.5 cm, or oedema

Proportion of children in sample fitting criterion for malnutrition

Monitor the trend of malnutrition in the community

PMC monitoring:
W/H of children between 6 and 59 months, consulting at curative clinic

<80% median W/H, or oedema, using the W/H chart

Proportion of malnourished children at curative clinic (PMC)

W/H = weight-for-height; MUAC = mid-upper arm circumference. a The pros and cons of W/H vs. MUAC, nor the merits and disadvantages of the W/H chart will be discussed in this text. b If age was not known exactly, as was often the case, children between 65 and 110 cm of height were included. Wherever is stated in the text 'children between 6 and 59 months' it means, in practice 'children between 65 and 110 cm'.

Monitoring of market prices

In early 1990, shortly after the arrival of the first refugees, MOH and MSF feared an imminent food shortage. Since prevalence of malnutrition is only a late indicator of a deteriorating nutritional situation, they tried to monitor market prices for the most common food commodities: rice and palm oil. This was in line with the Famine Early Warning Systems in use in drought-prone Sahel countries. In the Sahel, the ratio of grain price over cattle price - e.g. the price of 50 kg of millet over the price of a goat - has proven a useful indicator of food shortage. When food becomes scarce, cattle-owners start selling livestock and get decreasing quantities of grain in exchange.8,40 Inspired by this experience, MSF tried to monitor retail prices of rice and palm oil.

MSF officers visited the local markets in the refugee-affected areas on a weekly basis, but faced many practical difficulties to collect market prices that could be compared over time and between markets. Receptacles of different sizes and shapes were used as measuring units. The price might remain the same, but the size of one unit ('a cup') changed. The data showed a hectic evolution of prices. On a single day, for instance, palm oil could cost twice as much at one village market than at another; or prices could rise by 50% one week and drop again the next week. To what extent this instability of prices was real, or only reflected the practical difficulties faced by the officers who collected the data remains unclear. Despite the difficulties faced, it was clear that no systematic and sustained increase of food prices took place between January and May 1990, and this despite the fact that over 100,000 refugees had settled and no food had been distributed.

This stability of market prices was in sharp contrast with expectations. All remained convinced that a serious food crisis was forthcoming. The discrepancy between the perceived severity of the situation, and the apparent insensitivity of the market prices was explained as follows. Some thought that refugees did not purchase any food, but consumed exclusively food from Guinean household stocks. This would result in accelerated depletion of household stocks, but without increased demand on the market. It would then only be a question of time for major problems to arise. Others thought that the refugees actually faced food scarcity, but as they had neither cash nor livestock to exchange for food, there was, in economical terms, no demand.

SUBSISTENCE VS. MARKET ECONOMY. In actual fact, monitoring of market prices of commodities such as rice was probably not an appropriate tool in the Forest Region, which apparently has two parallel food economies: a subsistence food economy and a market food economy. This was definitely so for rice, maybe less for palm oil. Guinean subsistence farmers grow, store and consume their own rice, and have almost no livestock, or other marketable commodities that can be sold or exchanged in times of food shortage. When faced with a food shortage, people either limit food intake or borrow food, having then to reimburse 2 or 3 times the quantity after the next harvest.* Thus food shortage in families does not necessarily cause price increases on the markets. Rice markets in towns concern mainly imported polished rice sold by merchants to urban dwellers who do not farm. As soon as food aid was started in Guinea, market prices for rice collapsed, as part of the food aid was misappropriated and sold on the markets in the cities.

[* Food borrowing also became a common coping mechanism for refugees later, when food distributions were delayed. When finally food aid arrived, the indebted refugees had to hand over a large share of it to 'food lenders' and entered a vicious circle of indebtedness and impoverishment.]

Rural refugees integrated in the rural economy. Most took part in the subsistence farming system, somehow 'subsidised' by food aid, once this became available. Urban refugees, on the other hand, linked up predominantly with the market economy, where prices were very sensitive to food aid. Monitoring of market prices was thus not only practically difficult, but it also revealed only part of the reality and was not very relevant for the rural refugees. In the political discourse of UN agencies and donors, the low rice price on the market was considered as clear evidence for the harmful effects of food aid for local farmers. However, in the Forest Region, most farmers do not sell rice, but rely for cash on coffee and kola nuts. Low prices for rice were seen as harmful for local farmers, while this was never substantiated; it was more an indicator of the quantities that were misappropriated. The bottom line was that low market prices were a setback for those who misappropriated food aid and a welcome feature for those living in the towns, Guineans and refugees alike.

[ Prices for the prime cash crops, kola nuts and coffee, were never monitored, as they were not considered essential commodities. Kola nuts in the Forest Region are purchased by Mandingo merchants and transported to Upper Guinea, Mali, Senegal, and further north and east. It is likely that over 1990-96, due to the instability in the other production areas in Liberia and Sierra Leone, prices went up in Guinea. This possibly had a greater impact on farmers' income than the price of rice.]

The scarce data on market prices that were collected in Guinea showed that to interpret them several conditions have to be met. First, one needs a good understanding of the local economy, including knowledge of who is producing, importing, selling and purchasing which food commodities. Without these insights one falls quickly in the oversimplification that high prices for basic food commodities means difficult access to food for the poor, and that low prices mean a drawback for the farmers. Second, it is necessary to interpret market prices in the frame of the agricultural calendar of the area. Third, a simple indicator, similar to the one used during droughts in the Sahel (price of staple food/price of livestock) is not available (yet?) for a forest economy. In this context, it is not surprising that the monitoring of the prices of food commodities yielded inconclusive results, and the decision to stop it was probably correct.

PMC monitoring

WEIGHT-FOR-HEIGHT SCREENING. From the very onset of PARLS, weight-for-height (W/H) screening, using the Nabarro thinness chart, was introduced for children between 6 and 59 months consulting at the curative clinic.41,42 The nurse in charge of the curative clinic referred severely malnourished children (<70% median W/H, or oedema) for enrolment in a therapeutic feeding programme. Moderately malnourished children (70-79% median W/H) entered a supplementary feeding programme, and received a weekly take-home supplementary food ration.

MONITORING. MOH and MSF thought that the proportion of malnourished children at the curative clinic (PMC) could also serve as an indicator of the nutritional situation of the population in the catchment area of the health facility. They thought that using anthropometric surveys sparingly, guided by the results of routinely collected health service data (PMC) was a more cost-effective approach to nutritional monitoring than relying heavily on periodic anthropometric surveys.43 W/H of each child was thus registered on a tally sheet. At the end of each month, the proportion of malnourished children (<80% median W/H; brown or red on the Nabarro chart) among all children weighed was calculated. This yielded the PMC of the health facility. At the PARLS co-ordination offices, the PMC of all health facilities in one prefecture were pooled to calculate the PMC of the prefecture. The basic idea of PMC monitoring was that its trend would follow the trend of the nutritional situation of the community, even if it would overestimate the prevalence because of the 'vicious circle infectious diseases-malnutrition'.44-46

RESULTS. In some health facilities, the PMC remained stable over long periods. In many health facilities, however, monthly PMC followed a hectic course: e.g. Koundoutoh health post in Figure 21. Targeted anthropometric surveys in the catchment area of the health facilities showed that most steep increases in PMC were false alerts. These false alerts were due to chance occurrence, periodic active screening of thin children by the health animators, negligence during W/H measurements and calculation and transcription mistakes. PMC with pooled data were more consistent (Figure 21). PMC for N'ZkorYomou & Lola remained stable between 2.2% and 6.2% for almost 5 years. In Guu, PMC was always higher, and rose steeply during June and July in the 'famine season'.

Figure 21: PMC monitoring, 1990-96

USEFULNESS. PMC with pooled data yielded monthly information on the trend of malnutrition in the community. Such information could not be obtained through anthropometric surveys, as these were too resource consuming to be done at monthly intervals. It was necessary, however, to strictly standardise the data collection in the health facilities. Instructions to weigh all children consulting, even those visually well-nourished, and to weigh but not register for PMC monitoring thin children attending a vaccination session, did not easily fit in the patient-care logic of clinic staff. As PMC was the only indicator available monthly, it was sometimes used as 'the malnutrition rate'. This was clearly wrong. Moreover, PMC was not always higher than prevalence of malnutrition assessed by anthropometric surveys, as would, be expected. It was sometimes considerably lower. Registration of the absolute number of malnourished children consulting health facilities yielded similar information (Figure 22) and was less prone to errors. PMC monitoring made it possible to avoid a number of anthropometric surveys during 1990-94, when the situation was relatively stable. It did, however, not fully satisfy decision-makers who continued to request state-of-the-art anthropometric surveys.

Figure 22: Malnutrition at the curative clinic, Dieckealth centre, 1993-95

Anthropometric surveys

PARLS conducted many anthropometric surveys, with different sampling methods, anthropometric indicators, cut-off points (Table 11), and overall inconclusive results. The use of anthropometric surveys in PARLS changed over time.

1990: INITIAL ANTHROPOMETRIC SURVEYS. In early 1990, when despite repeated appeals the start of food aid was delayed, MOH and MSF thought food shortage was imminent. Although market prices and PMC monitoring did not show any worrying trends, these data could not calm the fears. MOH and MSF remained convinced that refugees were facing acute food stress and thought the monitoring systems were inadequate.

In March 1990, they therefore decided to carry out anthropometric surveys in the refugee-affected areas. They chose a cluster-sample method similar to the EPI survey method.47 90 clusters of 7 children each - 30 in each of the three sub-prefectures hosting refugees at that time - were surveyed, including refugees and Guineans. Weight and height were measured41 and the percentage of the median W/H of the NCHS/CDC/WHO reference population calculated for each child.26 The results showed that prevalence of malnutrition was 3.1% (95% CI: 1.4-4.8), without significant differences between the three areas surveyed, or between refugees and Guineans. All parties involved were surprised by this unexpectedly low prevalence.* MOH and MSF concluded that malnutrition had not yet increased, and that there definitely was no need for a supplementary feeding programme. Despite these findings, they remained convinced that family food stocks were decreasing, and that the nutritional situation would soon deteriorate.

[* In 1991, a nation-wide anthropometric survey found 4.7% of children <80% median W/H in the Forest Region, compared to 11.5% for the whole of Guinea. Stunting instead was relatively more frequent in the Forest Region (41.6% compared to 31.6% for the whole country).]

In May 1990, a CDC officer carried out a new anthropometric survey with 30 clusters of 30 children.48-50 This survey yielded a prevalence of malnutrition of 8.0% (95% CI: 5.5-10.5), again with no significant difference between refugees and Guineans. These results were interpreted as clear indication that the expected food shortage had finally started. MOH and MSF lobbied the different UN agencies to speed up their food aid. At that time, neither market prices nor PMC indicated a deterioration of the situation. The fact that CDC had calculated prevalence of malnutrition using <-2 z-score median W/H as cut-off point, whereas the previous prevalence was based on <80% median W/H, went unnoticed. The rates were not comparable. Re-analysis of the CDC data using <80% median W/H as cut-off point, gives a prevalence of only 5.3 % (95% CI: 3.2-7.4), which is not significantly different from the results of March 1990 (Table 12).

By July 1990, refugees received food aid on a regular basis. When WFP supplies food to refugees, they request implementing agencies to conduct three-monthly anthropometric surveys according to standard guidelines.26,51 EPINUT software52 is designed to analyse the survey data. In July 1990, MSF thus conducted a new survey in the same refugee-affected rural areas. It showed that prevalence of malnutrition was 5.3% (95% CI: 3.2-7.3), which was again not significantly different from the previous surveys (Table 12).

Table 12: Anthropometric surveys, March, - July 1990


Sampling method

Criterion for malnutrition


Prevalence of malnutrition


95% CI

March 1990

90 clusters of 7 children

<80% median W/H




May 1990

30 clusters of 30 children

<80% median W/H




July 1990

30 clusters of 30 children

<80% median W/H




a If calculated for <-2 z-score median W/H, the result is 8.0 % (95% CI: 5.5-10.5)

1990-92: RELIANCE ON PMC MONITORING RATHER THAN ON SURVEYS. After these surveys, MOH and MSF realised that anthropometric surveys consumed a lot of time and energy, without yielding conclusive evidence. When WFP and UNHCR again requested anthropometric surveys to provide them with 'the malnutrition rate of the refugees', MSF considered this inappropriate. Both MSF and MOH argued that one large anthropometric survey was useless as it would mask differences between the different types of refugees, rural and urban, and would be difficult to interpret. They resisted the pressure of WFP and UNHCR, but decided to continue PMC monitoring to detect trouble spots. They intended to perform surveys only when and where PMC would increase significantly. However, WFP and UNHCR did not easily accept this since they needed survey results for accountability reasons.

1992-95: INTAKE ANTHROPOMETRIC SURVEYS. During 1992-95, the small waves of late refugees arrived, and those were often in a poor general condition. Upon arrival, an exhaustive anthropometric survey was performed as part of the initial assessment. In the beginning, this was done using W/H, but as of 1995, surveys would use mid-up arm circumference (MUAC) as anthropometric indicator, with 12.5 cm as cut-off point. This faster method made it easier to include all children in a particular refugee group in a single day. Such surveys were often subject to selection bias, since parents often started bringing more children as soon as they noticed the survey activities.* Bias, accuracy and confidence intervals were considered to be of little importance, as prevalence of malnutrition was often very high indeed (Table 13), and food was urgently needed anyhow. What the NGOs and MOH wanted was an 'objective' criterion to be able to report statements such as 'malnutrition rate is over 20%', to convince decision-makers in WFP and UNHCR of the need to assist the new refugees as a matter of urgency. After such results, usually a first distribution of food could take place within only days.

[* Maybe they hoped that registration of their child on the health workers' list would entitle it to food rations. Community leaders maybe thought that a high prevalence of malnutrition in the community would encourage relief agencies to favour their community in any future distributions. Action Contre la Faim, however, used MUAC in a more systematic way, insisting to assess all children, and returning on the following day to measure children absent during the survey.

Table 13: Examples of small anthropometric surveys in newly arrived refugees, 1993-95



Sampling method

Criterion for malnutrition


Prevalence of malnutrition


95% CIa


Aug 1993


<80% med W/H





Nov 1993


<80% med W/H





Nov 1993


<80% med W/H





June 1994


<80% med W/H





Febr 1995


<80% med W/H





Febr 1995







Aug 1995


<80% med W/H





Aug 1995


<12.5 cm MUAC





Aug 1995


<12.5 cm MUAC




NA = not available; W/H = weight-for-height; MUAC = mid-upper arm circumference. a As original data were not always available, confidence intervals were not recalculated to verify accuracy. Confidence intervals were sometimes calculated, despite the fact that the survey was exhaustive.

1994-96: REVIVAL AND FALL OF LARGE ANTHROPOMETRIC SURVEYS. In 1994, pressure from WFP and UNHCR led MSF, Action Contre la Faim and MOH to conduct again a series of large cluster-sample surveys. WFP and UNHCR claimed to need survey results to plan quantities of food aid for the following year. MSF agreed, because they had doubts about the basic assumption of WFP and UNHCR that refugees registered in 1990-93 ('old' refugees) were less in need of food aid than those registered in 1994-95 ('new' refugees). The planning for future food aid needs was based on this assumption. Aid to 'old' refugees would be rapidly phased out, providing only food assistance to the vulnerable ones among them, while 'new' refugees would continue receiving food aid. In February 1995, anthropometric surveys showed that within geographical areas, no significant differences in nutritional status between 'new' and 'old refugees' existed.53 Beyond the initial period of a few months, time of arrival was not an important determinant of degree of self-sufficiency.54,55 Instead, prevalence of malnutrition among refugees living in real camps was considerably higher than among those not living in camps, and this independently from year of arrival.56,57 In Yomou prefecture, for instance, prevalence of malnutrition in Noonah camp was 14.4%, and in Bheta camp 14.7%, while in the rest of the prefecture it was only 4.3%. These results, however, did not influence the decision to distribute differential rations, based on year of registration: 'old' and 'new' refugees.

During 1995, quantities of food distributed were very low. This resulted in increasing PMC in the health facilities and a growing number of malnourished children in feeding programmes. UNHCR and WFP claimed to be in need of more objective information, meaning real surveys taking clusters over the whole refugee-affected area. MOH and MSF thus had to perform them. The results of some of these large surveys are shown in Table 14.

Table 14: Large anthropometric surveys, 1994-96 (not exhaustive)



Sampling method

Criterion for malnutrition


Prevalence of malnutrition


95% CI


Nov 1995


<-2 z-score med W/H





July 1996


<-2 z-score med W/H





July 1996


<-2 z-score med W/H




a Prevalence of malnutrition was 10.6% (95% CI: 7.9-14.0) among refugees and 6.1% (95% CI: 4.1-9.0) among Guineans.

Not surprisingly, the overall prevalence of malnutrition was not particularly worrying. It was obvious that the problems were concentrated in the small number of real refugee camps, and in areas with a very high concentration of refugees. The large anthropometric surveys could not easily detect this, as they resulted in an average prevalence of malnutrition for larger areas. Smaller, targeted surveys were carried out in the problem areas and these confirmed the existence of pockets with high prevalence of malnutrition (e.g. Fangamadou 12.6% and Koundoutoh 15.7%).

1996: PERIODIC EXHAUSTIVE ANTHROPOMETRIC SURVEYS IN SENTINEL SITES. When Action Contre la Faim, a specialised nutrition NGO, took over the nutritional programme from Mission Philafricaine in Macenta prefecture, they were not satisfied with the anthropometric data available. In 1996, they chose a few refugee settlements as sentinel sites and conducted bimonthly exhaustive 2-stage MUAC - W/H surveys of children between 6 and 59 months. This method yielded a precise prevalence of malnutrition in the sentinel sites. Soon, however, a bias was introduced. In the sentinel sites all malnourished children were identified, and were then, of course, included in feeding programmes. After two rounds, prevalence of malnutrition (<-2 z-score median W/H) in the sentinel sites fell below 2%. The sentinel sites were no longer representative of the rest of the prefecture. This method was resource-intensive, and did not yield the information it was set up for.

The limitations of anthropometric surveys

LARGE CROSS-SECTIONAL VS. SMALL TARGETED SURVEYS. The large anthropometric surveys were resource consuming, but rarely informative. One of the reasons was that most refugees could cope relatively well, even when food aid was delayed. Problems were concentrated in a few pockets. During large anthropometric surveys, these smaller problem areas were hidden in the overall picture. Smaller anthropometric surveys among newly arrived refugees, or in places where PMC monitoring had alerted to a possible problem were more informative. They often confirmed that prevalence of malnutrition was indeed high. Sometimes, however, they showed that PMC had given a false alarm.

SAMPLING PROBLEMS AND REPRESENTATIVENESS. In a complex and unstable situation, such as Guinea, it was difficult to determine the survey area. Which area was a refugee-affected one? One with at least 25% of refugees? Or should that proportion be higher? When the area had been defined, and all the villages and settlements in the area listed, determining the population of each of them was again problematic. Most often, the official registration data of UNHCR were used. But these were often over-estimations, and often refugees did not reside where they had registered. This was most pronounced in the refugee camps. Therefore, many children officially residing in these camps were not present during the survey, and were thus under-represented. Moreover, it seems unlikely that those present were representative. It could be that the poorest, and most malnourished remained in the camp. But it is equally possible that those facing acute food shortage and thus most malnourished moved out.

TECHNICAL PROBLEMS. In field situations, it is difficult to obtain reliable anthropometric measurements.* A more important problem, however, is that it is difficult to interpret results obtained with various indicators (W/H or MUAC) and cut-off points (<80% median W/H, and <-2 z-score median W/H). The so-called 'universal standard' for W/H is the NCHS/CDC/WHO reference population. In this reference population 2.5% of children are <-2 z-score median W/H, and these are defined as malnourished. There are, however, no universally accepted criteria for a cut-off prevalence of malnutrition in the community. The pros and cons of the different anthropometric indicators will probably remain a subject of debate over the decades to come.58-61 However, the use of both <80% median W/H and <-2 z-score median W/H as a cut-off for malnutrition is merely a matter of convention. The Nabarro chart, used at the curative clinic, is based on % median W/H. Z-score median W/H is more sound, as it is less dependent on age. The difference between both cut-offs depends on the age distribution of the children in the sample.62 In Guinea, <-2 z-score median W/H yielded prevalence rates some 50% higher than those calculated with <80% median W/H as cut-off (Table 15).

[* Weighing and measuring children seems simple, but in field circumstances, it is difficult to do it with the rigour required for reliable results. A particular problem was the determination of oedema. Pitting oedema is not easily standardised, and the a inter-observer reproducibility was low.]

Table 15: Use of different cut-offs for W/H during anthropometric surveys

Area & date


Prevalence of malnutrition

<-2 z-score %
(95% CI)

<80% median
% (95% CI)

z-score vs. %

Yomou, N'Zkoramp; Lola, May 1990


8.0 (5.5-10.5)

5.3 (3.2-7.4)


Macenta, July 1996





INTERPRETATION OF RESULTS. The confusion between indicators and cut-off points, added to the sampling biases and measurement errors, made it very difficult to interpret survey results, not least because of the difficulties to handle confidence intervals. Which is the indicated course of action if there is a large, but non-significant, increase in prevalence of malnutrition, from 5.2% (95% CI: 3.5-6.9) to 8.3% (95% CI: 6.6-10.0) over 3 months time? And what if, 3 months later it is reduced to 8.2% (95% CI: 6.5-9.9), although more refugees were absent during the survey, because it was performed during a period of intensive agricultural labour, which coincides with the famine season? If one considers moreover that many cross-border migrations in both directions continued over the years, results were truly very difficult to interpret.

PREVALENCE OF MALNUTRITION VS. FOOD SHORTAGE. Furthermore, given prevalence of malnutrition - or an increase in the prevalence - is difficult to interpret in terms of food shortage. First, there is usually a time lag between food shortage and the occurrence of malnutrition in young children. Second, within a family younger children may, or may not, get a higher share of the food available.63 Third, when food intake is limited during a longer period, it will result in stunting, and W/H or MUAC will not adequately reflect this. Fourth, when under-fives get supplementary food items, their nutritional status does not necessarily reflect the overall food availability anymore.

ANTHROPOMETRIC DATA AND DECISIONS ON FOOD AID. Finally, there existed serious doubts whether decisions on food aid actually considered the available anthropometric information. In mid-1990, WFP forced MOH and MSF to start a supplementary feeding programme (Box 4), despite the fact that the threshold of 10% prevalence of malnutrition was not reached, and that MOH, MSF, and even CDC had clearly advised to the contrary.* In 1995, UNHCR decided to phase out food aid for 'old' refugees and continue to supply 'new' refugees, despite anthropometric surveys showing that not time of arrival but settlement pattern and concentration of refugees were the principal determinants of high prevalence of malnutrition in certain communities. Most surveys did not show a significant difference in malnutrition between refugees and their hosts, but food aid was always exclusively for refugees.

[* In practice, if prevalence of malnutrition among young children is below 10%, it is often considered that there is no need for supplementary feeding programmes.6,26,48 It is unclear on which basis this cut-off point was defined. In Guinea, according to this criterion, a supplementary feeding programme was only warranted for some groups of late arrivals or in the camps (Table 13), but never for the general population (Table 12 & Table 14).

One could argue that precisely thanks to the food aid for refugees, the nutritional status of refugee children was similar to the nutritional status of Guinean children. However, even before food aid was distributed in mid-1990, no difference between refugees and Guineans was observed.]

A major issue right from the start

FOOD AID STARTED LATE. During the first months of 1990, food aid remained very limited. The French military flew some planeloads of food to N'Zkorand the Sierra Leonean government sent a few army trucks with rice. MSF obtained a budget from the European Union for local purchase of food to distribute to the refugees. When MSF tendered locally, the prices forwarded were considerably higher than current local market rates. MSF decided to withdraw from further involvement in food aid and the European Union decided to cancel local purchase of food for the refugees. UNHCR and WFP started the first significant food distributions in May 1990. Between 1990 and 1994 registration entitled refugees automatically to general food rations. From 1995 on, different rations were distributed to 'new' and 'old' refugees. Throughout PARLS, food aid to the refugees has always been the real stake. Medical care, water and education were much less important for the refugees, and for the Guinean authorities as well.

THE PROCESS. Food distributions were planned to take place every two months. Distributions were organised in dozens of distribution points within one or two days per prefecture. Refugees got a few days of advance warning through the refugee committees. ADRA transported the food from the warehouses to the distribution points, where the Red Cross, in collaboration with the refugee committees, distributed it. The military maintained order. The process of distribution involved calling a group of 50 refugees together in a fenced compound and collecting their cards. Officials of the Red Cross and the refugee committee took the cards, and checked them against the distribution lists. The members of the group then received their food, and took it out of the compound to divide it among them. UNHCR and WFP were supposed to monitor and control the operations, but the distribution system was left largely unmonitored for several years. Only from 1995 on, food basket monitoring teams* attempted to record fraud at the distribution sites. But refugee committees and the military sometimes refused them access to distributions.

[* Food basked monitoring consists of selecting a representative sample of the beneficiaries, and weighing, immediately after food distribution, the exact quantities of different food commodities received.26 ]

PROBLEMS WITH FOOD DISTRIBUTION. First, the refugees were generally not well informed about food distributions. Often, they were not given sufficient warning about the date of the next distribution. Refugees questioned in December 1995 often had no idea why 'old' and 'new' refugees received different quantities of food, what the plans were for future distributions and whether they would still receive food in the future. Second, the process of distribution was open to abuse. Members of agencies involved in food aid were aware of corruption, but felt incapable of addressing this. They also thought that levels of corruption were increasing over time. Eyewitness accounts of fraudulent events (Table 16) were widely discussed in the field and were consistently documented in all distribution sites.36,53 Fraud seemed well developed and institutionalised. Lastly, logistic capacity was a major constraint for efficient food distribution. Large quantities of food had to be moved within a very short space of time. ADRA had to rely on commercial vehicles, but there were no four-wheel drive commercial trucks available, so distribution was severely constrained during the rainy season, which coincides with the hunger season.

Table 16: Fraudulent events observed during food distributions

Some cards were served multiple times for a fee.
Some merchants held multiple cards and paid false beneficiaries to collect food.
Remaining stocks at the end of the day were reportedly distributed to the vulnerable, but in fact sold to merchants.
Oil tins were partially emptied and filled with water and distributed.
Food sacks were partially emptied and filled with stones and distributed.
The military forcibly took rations from some refugees.
Some refugees were refused food aid as refugee committees considered them to be 'rebels'.
Some refugee cards were 'blocked' and the ration was sold

On counting calories (1991-95)*

[* Data for 1990 could not be retraced, but were definitely lower than for 1991.]

QUALITY OF FOOD RATIONS. In 1990, the original food basket aimed to provide rice, oil, beans and dried fish. This was rapidly restricted to rice and oil only. In 1995, maize replaced rice as staple food. Over the years, roughly 90% of the quantity of food distributed was cereals, and 10% oil. Other items were much less important. Food aid distributed thus never supplied enough micronutrients. In 1990, this resulted in a small beriberi epidemic in the de facto refugee camp of Thuo in Lola prefecture (page 37). However, most refugees complemented relief food with vegetables and fruits, and did not face any vitamin deficiency.

Figure 23: Total food aid distributed, 1991-95

Figure 24: Kcal of food aid distributed per refugee, 1991-95

TOTAL QUANTITY OF FOOD AID. Figure 23 shows that the quantity of food aid increased considerably from 1991 to 1992. Food aid distributed in 1992 was calculated on inflated refugee numbers (the official data), and Guinean merchants misappropriated large quantities. After 1992, the total amount of food fell each year, despite an increase in the actual number of refugees (guesstimates). Therefore, between 1992 and 1995, the average quantity of food per refugee, based on the refugee guesstimates, decreased three-fold, and fell from 85% to 27% of the need of 2,100 kcal per day (Figure 24). UNHCR and WFP believed that the majority of refugees were self-sufficient and did not need food assistance.

1995: THE FAMINE YEAR. In 1995, operational agencies recognised a major increase in acute food stress and malnutrition.53 Between October 1994 and May 1995 no general food distributions took place. By May 1995, most refugees were in food debt and rations were often used to pay off debts. No further food distributions were made throughout the hunger season (May - September), contributing to acute food stress for many refugees. WFP blamed UNHCR for not having developed a better refugee registration, and WFP had major problems with the food pipeline.

FOOD NEED VS. PLANNED RATION. The official food basket for refugees in Guinea always provided less than the daily food requirements of 2,100 kcal per person per day. For 1995, the officially planned quantities of food provided 45% of caloric needs for 'old' refugees, and 85% for the 'new' refugees (Table 17). The planned ration thus clearly took into account that the refugees were partially food self-sufficient.

PLANNED RATION VS. DISTRIBUTED RATION. Because stocks were often not sufficient to distribute food as officially planned, WFP had to reduce the rations for distribution, and many distributions were cancelled or postponed. Only approximately one-third of planned quantities were effectively distributed, and this yielded only 16% of caloric needs for 'old' refugees and 28% for 'new' refugees (Table 17).

Table 17: Planned ration vs. distributed ration, 1995


Planned daily ration

Distributed daily ration

'old' refugees
grams (kcal)

'new' refugees
grams (kcal)

'old' refugees
grams (kcal)

'new' refugees
grams (kcal)


200 (720)

300 (1,080)


135 (486)


25 (221)


9 (76)

11 (99)

Corn Soya Blend


125 (475)




225 (941)

450 (1,776)

80 (330)

146 (585)

% of caloric need





DISTRIBUTED RATION VS. ACTUAL RATION RECEIVED. During the food distributions of September and December 1995, MSF performed food basket monitoring at the distribution sites.53 They showed that fraudulent practices reduced actual rations by on average 19% (range from 5 to 35%). The refugees thus actually received only 81% of quantities distributed.

Figure 25 illustrates the differences between daily food needs, planned rations, distributed rations and actual rations received by the refugees. 'Old' refugees received only 13% of daily subsistence requirements and new refugees 23%. Only some 40% of what was planned was indeed imported and distributed. Fraud during distribution accounted for a further 19% reduction in rations received. The caloric loss for 'old' refugees was 63 kcal per day due to fraud at distribution points and 611 kcal per day due to discrepancy between planned and distributed ration. For 'new' refugees it was respectively 111 kcal per day and 1,191 kcal per day. Although fraud accounted only for a minor part of the caloric deficit, it retained the attention of relief workers and donors, and influenced their attitude towards refugees and food distribution.

These data estimate the average quantity of food aid a registered refugee received during general food distributions. This average number of kcal per refugee still hides other problems. First, the amount received was probably even less for the weakest, as food was distributed to groups of 50, who then had to divide the rations between households. In practice, division between group members often resulted in fighting, and the most vulnerable being deprived of their share. Second, food borrowing was a common practice. After food distribution, families often had to give a large share of their food to people from whom they had borrowed food previously, often with 100% interest. Third, in 1995, WFP switched from distributing rice to maize flour. Refugees were not used to maize flour and did not like it. Consequently, many refugees sold the maize flour, on average at approximately FG100 a kilogram. Rice cost then between FG250 and FG500 per kilogram on the local markets. Therefore, 5 kilogram of maize flour was effectively traded for between 1 and 2 kilogram of rice.* The effective ration after sale of maize for rice left the refugees with a very small food income from food aid. Lastly, the unregistered refugees did not receive any food at the general food distribution. Fortunately, by 1995, alternative circuits for vulnerable refugees were starting up, where destitute refugees, particularly the unregistered, could obtain some food and cash.

[* In 1995, rice was more expensive on the international market, roughly twice the price of maize. However, the terms of trade for rice and maize flour at the local markets, and the refugees' preference for rice meant that the donors transferred less kcal to refugees, even if they had supplied only half the quantity of rice. If choice of the staple was purely a question of price, it would have been more cost-effective to supply rice.

These alternative circuits explain also the apparent contradiction between the 1995 data on food distributed of Figure 24 and Figure 25. The 27% in Figure 24 includes all food distributed. The 16% and 28% (weighed average 18%) in Figure 25 only include food distributed during the general food distribution.]

Figure 25: Food needs vs. food income from general food distribution, 1995

1996: Better supply and less fraud

In December 1995, a review of food aid in the Forest Region made all these problems explicit.33 According to UNHCR, the report increased pressure by donors to further decrease food aid to refugees in Guinea. In 1996, however, higher amounts of food were effectively distributed, mainly through less discrepancy between planned and distributed rations, but also because the 50-beneficiaries-group system was abolished and refugees received food per household. Food supply was thus more reliable in 1996 than in 1995. Food basket monitoring was performed systematically during all distributions, and results openly discussed at meetings with implementing agencies, UNHCR and WFP. According to the refugees, these measures considerably reduced problems and fraud during distributions. Eglise Protestante Evangque further developed its network of counselling centres and distributed food and cash to social cases. By mid-1996, they were distributing daily rations of 300 grams of cereals, 25 grams of oil and 50 grams of lentils (1,469 kcal per day) to 65,000 beneficiaries.

Other factors also converged to improve the situation. The political situation in Liberia had improved enough to enable again cross-border travel and farming in the home country. The 1995 famine had taught refugees that they could only count on themselves and they further engaged in farming in Guinea. UNHCR facilitated this, by establishing tripartite agreements between UNHCR, local communities and the refugees, to grow rice in the fertile but largely unused valley swamps. These improvements in 1996 were less pronounced in rural Guu, with its very high concentration of refugees, which reduced agricultural possibilities for them. Moreover, as Guu hosted mainly Sierra Leonean refugees, they could not re-establish links with their home country, where the political situation remained bleak.

Although there was a clear improvement in 1996 compared to 1995, the situation was still not satisfactory. For 1997, UNHCR made plans for more selective general food distribution. Only 'new' refugees and those considered vulnerable would still get food aid. A number of categories were proposed. These included: single parent households; unaccompanied minors, and families taking care of unaccompanied minors; families with a malnourished child; unaccompanied elderly and people with handicaps or special needs. There were, however, serious problems with the identification of refugees belonging to these categories. When refugees perceived additional benefits, they might simulate belonging to such categories. For instance, in a situation where family splitting was very common, it was easy for women to pretend to be the head of a family. Despite these anticipated difficulties, provisions were made for up to 30% vulnerable refugees among the total caseload: these would benefit from 300 grams of maize and 25 grams of oil per day (1301 kcal per day). Others proposed to limit food distribution in the rural areas to the traditional 'hunger season' during May-September. As a compensation for decreased food aid, investment in swamp farming would be increased, as well as school meals and food-for-work programmes.

Expert opinion versus anthropometry

In Guinea, rational decisions would have required a correct understanding of very different phenomena at various points in time. Such understanding appears to be a matter of 'expert opinion' rather than of 'hard anthropometric evidence'. Expert opinion should be understood as 'relevant qualitative information, which could be obtained through good field knowledge'. An analysis of eight key situations of relevance to decisions on food aid (Figure 26) shows the relative usefulness of various types of information for assessing the situation.

Figure 26: Insights in nutritional and food situation, 1990-96.

Fig 26

Description of the situation

Marginal usefulness of surveys as compared to expert opinion

1. Uncertainty

During 1990, MOH and MSF feared serious problems, but despite important delays in food aid, they did not occur. No objective assessment method indicated problems, but the conviction that trouble was soon to come remained predominant till food aid actually started. It is impossible to know whether this fear was justified - i.e. whether a famine would really have occurred without food aid.


2. Stable

Between 1990 and 1994, the overall nutritional situation remained stable. The number of malnourished children at clinics and hospitals, and PMC monitoring easily and convincingly indicated this.


3. Fraud

Between 1990 and 1994, serious problems with over-registration and misappropriation of food aid occurred. Details on quantities misappropriated and places where food was stored, individuals involved and bribes paid, were widely known and commented upon. Huge piles of donated food ('not to be sold or exchanged') were visible to everyone visiting the markets. The supplies depreciated the market price.


4. Late arrivals

The smaller waves of late arrivals often arrived in poor general condition. This was obvious during field visits, and small anthropometric surveys, whatever the criterion for malnutrition used, easily confirmed this.

Confirmation and advocacy.

5. Camps

The refugees who had to settle in camps remained very dependent on food aid. Numbers of malnourished children in these camps, PMC, complaints by refugees and their tendency to move out of the camps, clearly illustrated the problematic situation. Small anthropometric surveys in the camps could only confirm the obvious.


6. Famine

The 'story of an announced famine in Guinea, 1995' started with many clear qualitative signs of an impending problem, before the indicators started revealing it. Repeated alarms were given, but 'pipeline problems' and 'lack of commitment of the donors' prevented WFP from reacting adequately. Soon, all indicators - PMC, number of malnourished children, or anthropometric surveys - confirmed the existence of a famine in the relatively small proportion of refugees fully dependent on food aid. This famine could, however, not be tackled in time for the same reasons that impeded its prevention.

Late confirmation.

7. Maize

When in 1995, maize flour was distributed to the refugees, it was massively sold at a low price. Anyone present at a food distribution, visiting a market or travelling in the region could easily observe this. Many refugees sold the maize flour to traders immediately after the food distribution. Commercial trucks transported back the maize along the same dust roads over which relief trucks had brought it in the morning. No 'objective indicators' revealed this tragicomedy.


8. Improvement

In 1996, despite limited food aid, the situation remained satisfactory in most areas. These improvements in 1996 were less pronounced in rural Guu, with its very high concentration of refugees, and less cross- border movements. The number of malnourished children and PMC monitoring easily revealed this. Surveys confirmed it.

Confirmation and reassurance of relief workers.

Sound knowledge of the field situation in the Forest Region, through contacts with the various actors, and frequent field visits, was thus more useful than anthropometric indicators to understand the problems with food aid and lack of access to it. If and when useful, a simple indicator - as the number of malnourished children per month in a clinic or a hospital - revealed the problems as well as a more complex indicator such as PMC. Anthropometric surveys in a well-defined group of refugees confirmed the problems better than large surveys throughout the refugee-affected areas. Qualitative information was more useful, but still, lack of insight in the economic and social reality of the refugee-affected area impeded their sound interpretation.

Figure 27: Interference in decision making in food aid

Decision making in food aid

It is now clear that Figure 16, is an oversimplification of decision making in a real-life situation. Figure 27 shows some of the factors that interfere between technical estimation of food needs, however imperfect, and distribution of food aid to the refugees. Fraud and logistic constraints have already been discussed above. At times, UNHCR and WFP could not convince the donors to allocate funds for food aid already decided, even if the main donors participated in the joint assessment missions, and likely donor commitment was already taken into consideration during planning.

There are also other less visible factors at work that depend more on bureaucratic reflexes than on an analysis of the situation. Blanket food distribution to all refugees, for instance, whatever their situation or need, has remained a general principle for years, despite clear evidence that needs were unequally distributed across groups and over time. Food aid was scaled down after a few years, irrespective of the settlement pattern and the degree of self-reliance the refugees could develop. Hard evidence from surveys or from observation of the field situation apparently had less impact on decisions than the simple logic of the initial unfounded assumption.

These standard reflexes and simplistic assumptions seem part of a bureaucratic logic of large UN organisations. To take decisions that go against standard organisational policy and tradition is difficult in such environment, even more so, if most staff are on short assignments. These decisions are also firmly embedded in the political agendas and institutional interests of these organisations. Implementing NGOs, whose contact with the field might have allowed them to respond in a more flexible way, had little influence on these decisions. Some NGOs actively advocated more flexible approaches, others may have hesitated in going against the mainstream for fear of souring their relations with the government and UNHCR, their main source of funds.

Officially, PARLS approached food aid and malnutrition as technical problems, and did not put it enough in context. By supplying less than full food rations, WFP and UNHCR implicitly acknowledged that refugees in Guinea were integral part of the wider society and were partly self-reliant. However, the assessment of their nutritional situation was not performed with a system's view, but linked too simplistically anthropometric status with quantities of food aid distributed.

As elsewhere,25,64,65 political and managerial considerations determined decisions, rather than the technical arguments. The latter rather justified or confirmed the decisions. 'Smarter' relief approaches16,20,63 would (1) acknowledge the political nature of decisions on food aid, and (2) find other means than anthropometric surveys to provide a credible basis for rational decision making. This is not merely a question of effectiveness or efficiency of food aid, but also of avoiding negative effects on the host society. Humanitarian agencies should shift from 'distributing food aid to refugees' to 'improving food security for the vulnerable groups in refugee-affected areas'.1 This needs, of course, a system view, and a longer time perspective.