Sales of relief food
Lancet article on sales of relief food Marion Kelly 21.01.98
Barbara Reed and Jean-Pierre Habicht's Lancet paper Michael Golden 15.02.98
meeting in Geneva in autumn '88 Katherine Hilderbrand 17.02.98
1988 meeting Andre Briend 17.02.98


Date: Wed, 21 Jan 1998 13:55:59 +0000

Subject: Lancet article on sales of relief food


Dear all,

In case you haven't seen it, I'd like to call your attention to a very interesting

article: Sales of food aid as sign of distress, not excess, by Barbara

Reed and Jean-Pierre Habicht published in the Lancet, vol 351, 10 Jan 1998

pp 128-130. It deals with sales of relief food rations by refugees, the

refugees' reasons for these sales, and the effects of sales and subsequent

purchases on refugees' diets. It also draws out important implications for

policy and practice.

This is an excellent example of the kind of relief-related research and

sharing of lessons learned that I am always banging on about at meetings

and workshops. My congratulations to Barbara and Jean-Pierre (if they're

on the ngonut) and to everyone else whose assistance is acknowledged in

the article.



Date: Sun, 15 Feb 1998 12:37:26 +0000

From: Michael Golden <>

Subject: Barbara Reed and Jean-Pierre Habicht's Lancet paper


This paper by Barbara and Jean-Pierre has been forwarded by Rita Bhatia (UNHCR) to the list. It is very important and many of us do not have regular or up-to-date access to the lancet. I have "cut and pasted" it from the original attachment to reduce its length (by about half) and retyped the tables into ASCII, for those that cannot convert attachments and do not appreciate long messages. I am sure that we should discuss the implications of this paper which puts numbers upon what field workers have been saying for a long time and donors have strenuously denied.



(with statistical inferences)

Barbara A. Reed, M.S.,

Division of Nutritional Sciences, Cornell University

Jean-Pierre Habicht, MD, PhD,

Division of Nutritional Sciences, Cornell University


This is a copy of the article that appeared in The Lancet, Jan 1998, but also includes the statistical tests and inferences to support the statements made in the Lancet article.



In 1996 the sale and export of food aid from refugee camps near Uvira, Zaire, prompted a study of the use of food rations. However, research has revealed that sales did not reflect an excess of food in the camps. They were provoked by the absence of important components of the food basket, by cultural aversion to the staple (maize) and oil provided, by difficulties in food preparation, and by refugees' limited ability to diversify their diet and cover pressing non-food needs. Food sales improved the micronutrient content of refugee diets but at the expense of energy lost from an already energy-deficient diet. At most 23% of the refugee households were eating sufficient and nutritionally adequate diets. The poorest fifth of households were twice as likely to sell or exchange food as were other households and their diets were the worst. These findings demonstrate the perils of the gap between policy and practice in food aid distribution.



In February 1996 the World Food Programme (WFP) reduced food rations for refugees in Uvira, Zaire by 20% when visiting donors noted that food aid, primarily maize and oil, was being sold in local markets and even exported. The donors' interpretation was that the refugees were getting too much. Our study, initiated by the United Nations High Commissioner for Refugees (UNHCR) explored the underlying causes of food aid sales in Uvira.



The refugee population living in the eleven camps in Uvira, grew by 18% between March and August 1996, reaching a population size of nearly 200,000 refugees: two-thirds Burundian and one-third Rwandan. The daily food ration per person to be received from WFP for the Uvira camps in 1996, before the reduction, was 1936 kcal (350 g maize, 120 g pulse, 20 g oil, 30 g corn-soy blend [CSB], and 5 g salt). During the study period, after the reduction, the ration actually distributed varied between 1491 and 1594 kcal/person, generally consisting of: 270 g maize, 120 g pulse, 15 g oil, 30 g blended food and no salt. Maize was increased when there was no CSB, and oil and pulse calories were exchanged if one or the other was not delivered. The maize was primarily yellow maize in dried whole kernels. Occasionally white maize or maize flour was distributed. The pulse was a bean or, less often, lentils or peas. The oil was an unfortified, colourless, vegetable oil.

UNHCR was also to provide 200 g of soap per person per month. During the study period, about 1/3 of the planned quantity of soap was distributed.


Household survey

The study began with observation and open-ended interviews with key informants who were drawn from among refugees (especially women), Zairian camp workers, and Zairian farmers.

Refugees of a wide range of socioeconomic levels and age were selected to discuss foodpreferences, food acquisition, market purchases, food sales, problems in the camps, and income.

Later a survey of 1005 households, randomly selected across the camps in proportion to population size, was used to quantify the nature and prevalence of buying, selling, and eating patterns. These variables were compared across income groups using standard chi-square tests. Survey data were analyzed using the Center for Disease Control's EpiInfo software.

UNHCR provided a historical record of distributions of food and soap in the camps.

Both the qualitative inquiry and the survey instrument included cross checks for validity since refugee informants may respond to questions in ways they believe will solicit more aid.

To gain insight into factors that influenced households' abilities to cope, additional data were collected in 15 representative sectors (quartiers) across camps using a ranking exercise to characterize relative "well-being". In each sector refugee informants were asked to sort cards representing their neighbors' households into groups of similar "well-being" (using their own definition). They were then asked to describe characteristics of each group, including sources of income. Using what was learned from the rankings about the relationship of income sources to perceived well-being, the 1005 households surveyed were classified into 4 income groups according to what they reported as their most important income source.

To evaluate purchases reported, foods were divided into 5 categories: staples (major sources of energy: cereals, tubers, beans), complements (sugar, protein foods, or other nutrients not supplied by the food basket), salt, oil and non-specified (i.e., reported simply as "food"). Non-foods formed a sixth category.

Data collected by the household survey on "days without eating" in the previous week and the number of meals and foods eaten the day before were used to evaluate the quantitative sufficiency and nutritional adequacy of household diets. A diet was considered sufficient if the household had eaten 2 or more meals the day before and had eaten every day in the previous week. A diet was considered adequate if it contained at least one food from each of the following 3 categories: a low-protein energy source (oil, cereals, tubers); a protein source (beans, meat, CSB); and a micronutrient source (vegetables, CSB). This is a minimalist definition of nutritional adequacy, as there is no specification of quantity of food per person, and there are possible combinations that would not be adequate.



A household was defined as those eating together and sharing resources; generally they lived in a single shelter. The average household included 5 persons. 19% of households included no adult males; 52% of the population were under the age of 16. 90% of the refugees came from rural communities and had been farmers.


Food sales:

Maize, and less often, other commodities, were sold to obtain cash to buy other things, most commonly salt and soap, neither of which had been consistently distributed: Salt had been included in one-quarter of distributions during the previous 16 months (in none during the study period) and soap in one-fifth of distributions during the previous 2 years.

Maize was the first choice for sale. The dried corn was unfamiliar and not liked. In Rwanda and Burundi maize is eaten roasted fresh or as a flour blended with soy and sorghum in a weaning food. Cooked as whole grain, the maize required much fuel; hand pounding for use in porridge reduced fuel demands (about 90%, based on tests in Uvira) but the loss of volume left people feeling hungry. In tests conducted in the camps, losses from hand pounding of more than half were substantiated. Women complained that the maize caused diarrhea and that their children could not or would not eat it, and many parents of children in the therapeutic nutrition centers said their children had become malnourished because they refused to eat maize.

Parents sold maize to buy fruit, sugar, and cassava flour for their small children. Traditionally, young children in Rwanda and Burundi were first fed sorghum porridge with sugar and then mashed beans, sweet potatoes, bananas, papaya and avocado. In the camps, when possible, young children received as their first food CSB, the food aid commodity most acceptable to children for chewing, digestion, taste, and nutritional value. However, CSB had been distributed only twice during a 10-month period which ended during the study period.

In its absence, children were often given cassava porridge, the traditional Zairian weaning food and the cheapest flour on the market, but one of very poor nutritional value.

There were few mentions of sales of pulse or oil. Among these, the most frequent were sales of beans to buy lentils, peas or other staple that could be cooked more quickly with less fuel. That the firewood that was distributed in the camps met less than one-fifth the needs. What little wood there was in the environment was guarded by the local population for their own needs.

Sales of food for non-food purchases other than soap were rarely mentioned.

Refugees said they sold food aid to diversify their diet; they worked in the fields of Zairians to get non-food necessities.


Oil exchanges:

Refugees bartered vegetable oil for an equal quantity of locally produced palm oil, an exchange that was possible because the market price for the distributed oil was slightly higher than that of palm oil. Traders found it worthwhile to buy palm oil, exchange it for vegetable oil in the camps, and then sell that oil. Refugees preferred the taste of palm oil, particularly to make maize dishes more palatable.

Extent of food aid sales and exchanges: Responses to questions about households' most recent food aid sales and the frequency with which they exchanged oil showed that 42% of households reported some food aid sale, and that 80% of these (33% of the total) had most recently sold maize. 54% of the households had exchanged their oil at least once. Only 30% had neither sold nor exchanged food aid.

Reasons for sales given in the survey accorded with those given by the key informants. 49% of sales were to purchase another food: because they preferred other types of food, to break the monotony, or to balance their diet. Some identified specific dietary needs, such as, foods "rich in iron"; anemia was prevalent in the camps. 25% of sales were made to buy salt. 12% of sales were because of intolerance to a commodity, usually oil (56%) or maize (41%). Only 9% sales were made to purchase non-food necessities, mostly soap (53%), clothing (25%), or petrol for tin-can lights (15%). 5% of sales were to buy medicine when none was available from camp medical services.

Soap was distributed near the time of the survey, so sales for its purchase were not as prominent as they had been during the interview period. In contrast to the interviews, the survey elicited no explicit responses about sales to buy food for children; however, households with at least one child (< 16y) were 40% more likely to have sold a food aid commodity than households with no children (p < .01).


Use of sale proceeds:

The reasons given for selling food were cross-checked with answers to a question about the sources of money used for recent market purchases (table 1). 85% of the purchases made with food aid proceeds were food, and salt (35%) and complements (35%) accounted for much of these food purchases. Funds from other sources were used to buy more non-food purchases and staple foods and less salt (p < .05).


Diet sufficiency and adequacy:

Key informants reported that the food distributed for 15-day periods lasted only 2 to 10 days, depending on the specific commodity and household size. Maize lasted 8-10 days; oil and CSB were finished in 1-4 days.

In the week preceding the survey, 45.5% of the households had gone without eating for at least one day. Usually (84%) because they had no food. Another strong indication of food shortage was that beans (including lentils or peas), the favorite food, were eaten the previous day in only 57% of the households.

The refugees' diet was, at best, both sufficient and adequate for less than one-quarter of the households (23%); for nearly half (49%) it was neither sufficient nor adequate (table 2).



Household food consumption the day before the survey and eating frequency data were compared across income groups (table 2), and differences (p < .05) were revealed. The four income groups used, in order of decreasing income were: (I) those with salaries or for whom commerce was most important, (II) those working in construction jobs or providing transport (bicycle) and dairy farmers, (III) those doing field work for Zairian farmers, (IV) those collecting natural products (grass, reeds, brush) for sale, doing household tasks for Zairians, depending most on others for money, depending most on food aid sales, and those with no income. With decreasing income, the average number of days not eating in the preceding week increased, while the number of meals and both the adequacy and sufficiency of household diet decreased (table 2).

Other findings (p < .05) were that the households with the lowest income were twice as likely to have sold or exchanged food aid proceeds than were other groups, and that non-food articles represented 18% of recent purchases made by the lowest income group, less than in the other groups (23.5-26.5%).



In Uvira, refugees were selling food aid commodities to satisfy basic dietary and non-food needs, most of which would have been satisfied if relief targets for distributions had been met. They were not selling because they had more food than they needed.

A quarter of the food aid sales reported would have been avoided with delivery of promised salt.

The absence of salt is not evident when food aid distribution is judged in terms of daily energy content. However, in Uvira, the energy cost of maize (0.5 kg) sold to buy salt for 1 day for a family of 5 (25 g) was 350 kcal/person, which is 18% of the total target daily ration's energy content almost 25% of what was actually delivered.

The most common non-food purchase, a bar of soap, when used to wash people, dishes, and clothing, lasted 2 days, on average, and half a bar of soap was equivalent to the cost of 1.5 kg maize, the entire day's maize ration for 5 before food aid reduction, or 1050 kcal/person. When soap was not distributed, it was used less and rampant scabies ensued.

Basic food aid commodities provided by through WFP are not intended to meet all nutritional requirements and UNHCR is responsible for seeing that other dietary needs are met. (1) In Uvira no meat, fish, or fresh fruit and vegetables were distributed. 60% of refugee households had cultivated complementary foods but this was inadequate, at least during the dry season. So, the primary means of obtaining complementary foods was by purchase. These refugees knew about good dietary habits, but less than a quarter had a diet that might have been balanced and met energy requirements, even by minimalist standards.

Purchases of basic non-food items (clothing, cooking fuel and materials, bedding, shelter materials), which had been distributed in insufficient quantities or not at all in Uvira, were less often financed through food-aid sales, but they diverted earned income that could have been used to purchase food.

The glut of maize on the market and in the port was clearly due to its low acceptability to the beneficiary population. Refugees sold maize even though, because Zairians do not like it either, it brought the lowest return per calorie. Sales of beans or oil to buy salt would have resulted in losses of fewer calories. WFP has known since 1992 that Rwandans do not like maize (3), and WFP/UNHCR policy is to distribute foods that are culturally acceptable (1) -- yet they not only continued to accept maize from donors anxious to unload surpluses and WFP even bought maize in Africa for distribution to these refugees, because no other cereal was available at a comparable price per unit of energy. Maize would have been more acceptable if milled, as was WFP policy (1), but WFP/UNHCR were not successful in setting this up in Uvira.

Given that relief efforts will never wholly satisfy refugee needs directly, refugee must acquire necessities by other means. Instead of reacting negatively to refugee trading in donated food ties, relief organizations could have accepted this as a legitimate coping mechanism and responded by facilitating the market. In Uvira maize was a poor bargain; it sold for only about 20% of its cost to WFP. Oil, CSB, and rice sold for just about the same as their costs to WFP and sugar and salt prices in the Zairian market were three times their costs to WFP. The provision of salt and the substitution of any of these other commodities for maize would have permitted refugees to buy more. The bulk of food aid oil in the Uvira markets arrived through exchange, not sale. Refugees preferred palm oil. Furthermore, most of the oil distributed was from the USA and was not fortified, so replacement by palm oil added vitamin A.


Implications for aid policy

Our finding that the appearance of food-aid commodities in local markets cannot be attributed to food excess is not unique to Uvira (2, 4-6) The reasons behind food aid sales must be explored before rations are reduced. In Uvira, the reduction in the rations merely exacerbated problems originating from inefficient international relief efforts.

Energy content is the yardstick most commonly used to judge ration sufficiency but our survey shows that neglect of cultural preference, micronutrient needs, and food preparation can seriously reduce energy intake. It reinforces the wisdom of existing policies that mandate the delivery of culturally acceptable foods (including non-caloric items), the provision of cereals in milled form and access by refugees to complementary foods. UN agencies and donors should take heed that these policies are acted upon.

The appearance of oil on the local markets of Uvira, had it been properly understood, should have heartened those responsible for the refugees because it was evidence of successful, self-sustaining, cost-free action on the part of refugees that resulted in an improved diet.

Less heartening is that more than three-quarters of households did not have a sufficient and adequate diet -- and most disheartening is the fact that rations were reduced because food sales had been misinterpreted by donors and responded to inappropriately by WFP.


Table 1 Comparison of purchases made with proceeds of food-aid sales and those made with other funds

Type of purchase % of purchased items
  Funded by food-aid sale not funded by food-aid sales
non-food 15.0 23.9
staples 24.1 28.5
complements 30.0 24.8
salt 29.6 20.6
oil 1.0 1.0
other food 0.7 1.1
Total food 85.0 76.1

Table 2 Adequacy and sufficiency of diets of differetn income groups

earning group (%) Days without eating no of meals previous day

previous day's diet

      sufficient and adequate insufficient and inadequate
I 8.4 0.3 2.0 55% 26%
II 5.0 0.5 1.9 35% 39%
III 66.4 0.9 1.7 21% 49%
IV 20.1 1.1 1.6 15% 60%
ALL 100.0 0.9 1.7 23% 49%


1. World Food Programme, United Nations High Commissioner for, Refugees. Memorandum of Understanding on the Joint Working Arrangements for Refugee, Returnee and Internally Displaced Persons Feeding Operations. Geneva, Rome: UNHCR/WFP 1994.

2. Hansch S. Diet and ration use in Central American refugee camps. J Ref Stud 1992; 5:300-312.

3. Shoham J. Food aid and nutritional programmes during the Rwandan emergency. Disasters 1996; 20:338-352.

4. Christensen H. Survival strategies for and by camp refugees, Report 82.3. Geneva: United Nations Research Institute for Social Development 1982: 1-48.

5. Harrel-Bond B, Voutira E, Lopold M. Counting the refugees: Gifts, givers, patrons and clients. J Ref Stud 1992; 5:205-225.

6. Wilson KB. Enhancing refugees' own food acquisition strategies. J Ref Stud 1992; 5:226-246.



It was Robert Saaf (UNHCR Uvira suboffice) whose idea spawned the study, and we thank him, Josiane Ntanyosha, and other UNHCR international and national staff in Uvira for their encouragement and assistance in the camps; the refugees for their participation; to members of the UNHCR/Geneva Programme and Technical Support Section, especially Rita Bhatia and Arnold Timmer, Alphonse Malanda (UNHCR Sub-office in Uvira at the end of the study, and Dieter Hannusch of WFP for their help and advice.

We wish also to show appreciation to Gretel Pelto and Betty Kirkwood, who reviewed and commented on the final drafts.

Financial support was provided by UNHCR Programmes and Technical Support Section (PTSS), the UNHCR Special Operations Zaire Desk, and Cornell University.


Prof. Michael H.N.Golden

From: "Katherine Hilderbrand" <>

Date: Tue, 17 Feb 1998 09:46:27 +0000

Subject: Ngonut: meeting in Geneva in autumn '88


Dear ngonuts,

In order to respond to the article in the Lancet which Prof Golden forwarded, I am looking for the recommendations of a meeting in Geneva at WHO which I attended in '88. It was a meeting of agencies working in food aid in emergencies IO's and NGO's and as far as I can remember was organised by Jeremy Shoham and WFP. At the meeting it was agreed that minimum rations should be 1.900 kcal AND that refugees in camps and self-settled refugees could sell rations to diversify their diet and meet non-food needs.

Does anyone have any record of this ? If yes, could the relevant portions be sent through the network ?

Thanking you in advance


Katherine Hilderbrand

Date: Tue, 17 Feb 1998 12:38:36 +0100 (MET)

From: (Andre' BRIEND)

Subject: Ngonut: 1988 meeting


Dear NGO nuts,

I guess Katherine is refering to the following report:

Nutrition in times of disasters. Report of an international conference held at the WHO headquarters, Geneva, September 27-30 1988.

I scanned p 21 para 4 :

4. Deficiencies of Other Nutrients

When populations are wholly dependent on a diet of only cereals, oil, and dried skimmed milk, deficiencies of vitamin A, B-complex, and C, iron, and folic acid develop. The provision of high-dose vitamin A supplement to children under five years old is a common and widely accepted practice.

Epidemics of scurvy associated with low dietary vitamin C have occurred recently in refugee populations in Sudan and Somalia. Attempts to manage the problem through distribution of vitamin C tablets and use of limes have not been successful because of problems in distribution, compliance, and procurement. The obvious solution is either to provide a nutritionally full and adequate ration or to permit trade that will allow aid recipients to obtain other foods to make up an adequate ration. Trading for spices and condiments should be permitted to break the monotony of the diet and to improve the palatability of donated foods. The authors see no reason to resort to distribution of nutrients in medicinal form, except for vitamin A.

There are no technological constraints to nutrient fortification of relief commodities, and fortification should be considered for populations entirely dependent on distributed foods.

This shows that unfortunately, excellent recommendations done during meetings are often forgotten. Hopefully, the Lancet article will be more influential that this forgotten meeting.

The 1988 meeting did indeed recommend 1900 kcal as working figure. I scanned the relevant recommendation:

The 1985 FAO/WHO/UNU energy and protein requirements should be used as the guide for calculating ration levels, especially if the population is totally dependent on distributed foods. If it is not possible to meet these requirements, then a minimum of 1,900 kilocalories per day per person should be the target for sedentary populations. Additional allowances must be added for non sedentary groups, groups at risk, and groups exposed to severe temperatures. The protein content should be at least 12 percent of the total calories provided. The rations must also provide for minimum requirements of vitamins A, B, and C, iron, and folic acid.

Please note that this carefully worded statement lead to much confusion. The figure of 1900 kcal was replaced by 2100 in the new agreement signed between WFP and UNHCR (MOU, 19th December 1997, Rita has more details).

Unfortunately, the 1900 kcal magic figure was rarely achieved in the years following the 1988 meeting.


Dr. Andre' Briend