Distributing food (Special Supplement 1)

Food may be distributed in many different ways but the method of distribution will, to a large extent, depend on the eligible groups and the method for identifying them. Distribution points may be developed using existing buildings, e.g. health centres, schools. Food distribution centres may be designed and constructed by agencies, e.g. agency developed supplementary feeding centres or distribution points for general rations. Food may be provided directly through home-based care, community kitchens, or given directly in bulk to the community for redistribution. For example, supplementary rations may be given to children in a supplementary centre to be consumed in the centre, to the family with directions that the ration should be given to a particular child, or to the community for distribution to children meeting agreed criteria of need. A therapeutic feeding centre providing high quality care can be built and staffed by the agency managing the distribution, and a network of distribution points and mobile teams, including community workers, might be used in support.

In order to support the effective targeting of food aid, there must be a distribution system in place which:

• Maximises coverage of the eligible group, and
• Minimises inclusion errors, i.e. keeps the number and proportion of non-eligible people who receive food to a reasonable minimum.

In situations where such systems are absent, targeting is likely to fail. Box 12 describes the failure of targeting in Darfur, Sudan in 1985. The reasons highlighted in the box demonstrate the weaknesses in the distribution system, which allowed diversion of food aid and prevented adequate coverage from being achieved.

Designing food distribution systems to maximise coverage

The chief factors contributing to low uptake of services are:

• A lack of awareness that a distribution is taking place and/or the eligibility criteria, e.g. recipients who meet the criteria for receiving food, but are unaware of this. This results in under-registration of intended beneficiaries. Figure 4 shows the improved coverage achieved after a communication campaign about a feeding programme in Malawi. Good communication strategies can also minimise exclusion errors, which can occur when food is provided for individuals who are not registered, but who appear at distributions in the hope of receiving food.
• The cost to the beneficiary of getting the food. The cost is the 'opportunity cost' of participation, i.e. the value to the beneficiary of attending a distribution, relative to the alternative to attending the distribution. The costs include travel and waiting time, occasional direct costs (e.g. fares), and the cost of transporting the food. Costs will increase where carers are expected to bring small children to centres, to attend frequently, or to spend long periods receiving care. The value of the service will fall if the service quality and cultural acceptability of the programme is low, e.g. crowded dirty centres. The alternatives to attendance at a food distribution may include cultivation, paid and domestic work, which may have to be done to ensure household survival, e.g. water collection.

Coverage can be improved by reducing the distance people have to travel to distribution points and the frequency with which they have to attend. In general, there is a trade-off between the need to bring food relief close to the beneficiaries, e.g. distributing food at village level, and the cost and administrative difficulty of doing so. However, the agent that bears this cost should be considered. A relatively small increase in costs for the agency may realise big cost reductions for the recipients (Valid International, 2003).

Distributing large amounts of food can make it hard to transport food

The trade-offs made in programmes that distribute through multiple decentralised distribution points, and those which focus on fewer distribution points, are well illustrated when comparing centre based and community based therapeutic and supplementary feeding.

Centre based feeding requires high levels of human and financial resources, and tends to have to high costs per individual beneficiary. The number of centres that can be supported is usually limited. In most cases, health care is also provided. Inpatient therapeutic feeding programmes can, sometimes, achieve mortality rates below 5% and weight gain greater than 10g/kg per day. Both centre based therapeutic and supplementary feeding often suffer from high rates of patient defaulting. The practical problem arises chiefly where beneficiaries are often expected to attend daily, and where they may be required to remain at the centre for some time. An extreme example was found in Ruhingeri, Rwanda in 2000, where a woman was required to walk an 11 km return journey, 6 days a week with her child, to receive a small supplementary ration. The energy content of the ration approximated the energy expended on collecting it, and a visit to the centre required the woman to forego a day of paid work or obligatory communal work.

Community based feeding tends to have slightly lower costs per beneficiary, and allows many more distribution centres to be run although logistic demands are high. Treatment protocols tend to be less sophisticated though health care is usually also provided. The recent availability of ready to use therapeutic foods have made community based therapeutic feeding possible. Even though therapy is less intensive in community based therapeutic care and consequently weight gain is lower than centre based care, mortality rates as low (<3%) (Taylor, 2002) as those achieved in centre based care have been reported probably because the risks of cross infection are minimised. In practice, community based feeding programmes do require a small centre based component for treating severely malnourished children who are acutely ill, but usually the scale of requirement for this is reduced as children present earlier for treatment.

Figures 5 and 6 illustrate the coverage achieved in centre based, as opposed to community based, feeding. Figure 5 compares the location of distribution points for community based therapeutic care (Outpatient Therapeutic Programme (OTP)), compared to those for centre based care (therapeutic feeding centres) in a famine relief programme in North Darfur, Sudan in 2002. It would be anticipated that greater coverage could be achieved through the community-based care, simply due to the sheer number and geographical coverage of the distribution points. Figure 6 compares two districts in Malawi in 2003, Mchinji and Dowa. In Mchinji district, centre based therapeutic care was implemented in Ministry of Health facilities with support from Save the Children UK. In Dowa, Concern introduced community based care. The size of the shaded squares is directly proportional to the coverage rate found in the village surveyed. Villages were selected for survey in order to provide a geographic spread across the district. It can be seen that considerably greater levels of coverage were achieved in Dowa.

Practical difficulties faced by targeted beneficiaries, related to the frequency and organisation of the distribution, should be taken into account. For example, older people may not be able to queue for long periods and have difficulty in obtaining access to information about their entitlements (HelpAge International, 2001). Several examples illustrate the problems of distribution design and how it can undermine targeting.

• In Zambia, women working on a road construction project were required to carry headstones from a quarry to the construction sites. This continued until the distances became so great that they had to hire men with carts to assist them, in exchange for 50% of their wages (Devereaux, 2000).
• If large amounts of food are distributed in order to minimise the frequency of distribution, people may have difficulties in transporting the food. For example, the Ethiopian government distributed large quantities (sacks) of grain to people in Korem, in the run up to the 1984/1985 famine in the hope that this would avoid the creation of a camp. In fact, people could not always carry the food, or feared robbery on the road, so that the very effect that the authorities hoped to avoid occurred, with people settling on the periphery of the town.
• In the Ethiopian Ogaden in the 1980s, a large effort was made by an organisation to distribute food selectively to women and children in a rural pastoral population, using a town based food distribution system. The food was promptly re-bagged by the recipients in order that it could be transported.

Designing food distribution to minimise inclusion errors

Inclusion errors arise principally from:

• Poor application of eligibility criteria
• Diversion of food by the recipient, community or people involved in the distribution chain

Various means are used for identifying the eligible population including registration, the use of bracelets, ration cards and skin dye. In centre based feeding and food distribution programmes, inclusion errors can generally be kept to a low level if eligibility criteria are objective, staff are clear about the eligibility criteria and adequate supervision checks are in place. Theft and fraud can be minimised, although rarely entirely eliminated, by ensuring that record keeping and supervision procedures are in place. If eligibility criteria are not clear and cannot be easily applied, then inclusion errors will result (see Box 5). Targeting guidelines can assist with the implementation of eligibility criteria at distribution points.

Inclusion errors do arise in centralised systems, in most cases because the food available is less than that required (see Section 6). Under these conditions, staff may come under extreme pressure to include beneficiaries who are not eligible. However, even in these contexts, programmes with strict criteria can result in minimal inclusion errors. For example, a re-registration of all children registered in the Save the Children UK supplementary feeding programme in Huambo, Angola in 2000 (see Box 10) was conducted. The lack of food support to the general population led to enormous pressures on feeding centre staff to admit children. At that time, 9600 children were registered in the centres and only 4.7% of these children were found to have met the exit criteria, i.e. were no longer eligible (Save the Children, 2001).

In many instances, particularly when households rather than individuals are targeted, there will be a degree of diversion of food aid in the system, resulting in inclusion and exclusion error (see Box 13). Box 13 also shows that the distribution system must be designed with sensitivity to the political situation and the risks of diversion (see also Section 2). The extent of diversion will vary, according to the context and the targeting system. While the prevention of deliberate abuse is a major preoccupation in many targeting systems, there may be other more important sources of error in many targeting systems (such as the design of the system).

There is a growing body of experience demonstrating methods for reducing diversion in situations where communities, themselves, have greater control over the food aid resources (see also Section 5) and distribution is decentralised. The success of these experiences appears to lie first in the suitability of the context, and secondly, in the attention paid to the process of developing the targeting system and in ensuring it is transparent, and that all those involved feel a degree of accountability to one another.

Community managed targeting (CMT) places a greater responsibility on the community to decide who requires food assistance and ensuring they receive it. CMT aims to achieve the transfer of responsibility for targeting, managing and monitoring the food distribution from the implementing agency, to the beneficiary community (Mathys, 2003). Figure 7 shows the key principles on which the approach is based, and Box 14 shows the key steps recommended for effective community based targeting.

There have been a number of reviews (Jaspars et al, 1997, Ridout 1999, Jaspars and Shoham, 1999, Shoham, 1999) of community based targeting which have shown that this approach is only likely to be feasible in certain conditions, where:

• All key stakeholders (from community to national level, and implementing agencies) share common objectives concerning targeting and participation, and where parts of the population at risk are not politically marginalised. For example, many people displaced from Wau into Bahr Al Gazal, Sudan during the 1998 crisis, were excluded from distribution as they were not regarded by the local community as legitimate recipients of food.
• There are cohesive social groupings living in peace and stability, where recipient groups are smaller, are clearly geographically demarcated, are related and are economically interdependent.
• The emergency has not reached crisis proportions, or rates of malnutrition and mortality have not become excessive, and where food aid is targeted at the majority of the population (so that inclusion errors can be accommodated without negating the targeting process).

Women selling food in South Sudan

There is a broad sense that it is more practical to target reliably, and with least 'slippage', with settled than with pastoral peoples (Acacia, 2002). In pastoral populations, community relationships and obligations may exclude the possibility of targeting to specific households, i.e. the food may simply be shared amongst all. Although it is conceivable that certain protracted refugee situations may comply with the conditions identified above, community based targeting may often not be appropriate in refugee contexts. Specifically, refugees may be used to the general provision of a full ration and therefore, be unwilling to participate in a process of targeting. The high frequency of distributions in many refugee situations would pose considerable time burdens on the Relief Committees, and the strategy may not work well for new arrivals (UNHCR et al, 2000).

Involving women in food distributions can improve targeting

Aspects of community managed targeting have been applied in conditions, other than those above, with mixed results (see Box 15). Deviation from the eligibility criteria and frank diversion of food by powerful groups in the community can result, if the context is not suitable. For example, villages may be excluded by the Relief Committee or chiefs due to distance, ethnic differences or political marginalisation, while households may be included because they are related to, or politically aligned to Relief Committee members or chiefs. Similarly, households that were politically marginalised may be under registered, i.e. fewer household members than actually exist may be registered. In Zimbabwe, Save the Children UK has shown that the inclusion error increased dramatically over 3 years of targeted food aid distributions (see Table 4) as the economic and political context changed.

Food drop in South Sudan

Involving women in food distributions is another aspect of community participation which can improve targeting. Experience shows that in situations where women are the household food managers, they are more likely to ensure that food resources reach the children, and women are less likely to sell them. In addition, rations intended for polygamous households, given to the male household head, might not be equally redistributed between wives' families. Targeting women with food assistance is a requirement of WFP assisted food aid programmes. In practice, this means ensuring that women control the family food aid entitlement in 80 percent of WFP handled and subcontracted operations, which target relief food distributions to households. In some situations, efforts to reduce discrimination against women in the design of targeting programmes have been shown to be successful (Acacia, 2002, Chapman, 1998a and WFP, 2001).

Box 16 shows how including women in the process of decision making for food aid allocations in Southern Sudan increased the likelihood that food reached those most in need. This was possible, in this case, because the social context and the roles and responsibilities of different community actors were understood in depth (see Box 16). Targeting food aid in this way requires re-negotiation of social roles, which takes time and involves the risk of alienating certain groups and inciting conflict. In locations in southern Sudan where this negotiation was not done, the system was regarded locally as having been imposed and the chiefs redistributed the food immediately after the distribution (see also Section 6).