Resourcing Targeting (Special Supplement 1)

Agencies deciding to implement food aid targeting in emergencies often face constraints on targeting design. Three such factors are dealt with here:

• Quantity and quality of food typically available for emergency food aid targeting
• Overall quantity of resources available to respond to a particular emergency
• Quantity of resources required to put in place an effective targeting system (i.e. the costs over and above the food costs)

These three factors heavily influence the feasibility of the design of any targeting system.

The quantity and quality of food required

Typically, a 'general ration' in a poor developing country is made up of cereals, sometimes pulses, less often oil and very occasionally, salt and other items. Calculations of food aid needed for households are usually based on an average population nutrient requirement which includes the additional needs of pregnant and lactating women. Even recommended rations fall short in some aspects of dietary quality (see Box 21). Ration planning should, therefore, also include recognition of any special food commodities required to meet the needs of physiologically vulnerable groups in the community, such as the people living with HIV/AIDS, the sick or elderly.

However, in spite of requirements and recommended rations the size and composition of the ration available to be given to emergency affected people varies widely, largely according to the availability of specific commodities. In some locations, the ration falls far short of requirements while in others, e.g. the Balkans, more elaborate rations have been distributed and elsewhere, e.g. Iraq, rations given with the wider objective of income support may include non-food items, such as soap and detergent. General rations are often inadequate, in either quantity or quality, to meet the needs of all individuals. This makes a case for targeting subgroups of the population who have elevated requirements for nutrients (see Section 1).

Targeting under conditions where the food available is less than needs

Situations where the quantity of food aid available is much less than the assessed survival need, i.e. where there is no possibility of providing aid to meet household needs, however these are defined, have regularly occurred. In this situation, there are two choices, either to give everyone a tiny share, or target the available resources at a particular group facing the greatest risk. The decision will depend on the context and also the future estimate of availability of resources. For example, in an acute emergency and before the food pipeline has been established and more discriminating systems organised, an agency may decide to do a blanket food distribution for all children aged under five years. This will provide a meagre household ration and may help to protect those most vulnerable to malnutrition, i.e. under fives. The commodity selected may be chosen with children's consumption in mind, e.g. a blended food, or blended food pre-mixed with sugar and oil, in the hope that this will strengthen the targeting by making it difficult to include this in the general household diet.

In the medium term, the management of a situation where there is a gross shortfall in the amount of resources available presents difficulties. Assuming that, a) a population has no alternative lines of action open to them, e.g. migration, b) the food available, even if equally distributed in proportion to need, is insufficient to ensure long term survival, and c) it is not anticipated that the shortage will be quickly relieved e.g. by a harvest, the outcome must inevitably be the death of some proportion of the population.

In designing a system to deal with this situation, a reasonable aim is to ensure the survival of the maximum proportion of the population for the longest period. A logical approach is to provide a ration to the population at greatest risk of death, sufficient to ensure recovery to a level which will reduce that risk, e.g. to provide food only to children 80% of the median weight for height and other age groups, to the extent that risk can be assessed, until they have recovered to the point where risk has been reduced (Seaman & Rivers, 1988).

In practice, the systems which have been used broadly follow this logic, but in many cases, modify the eligibility criteria to include carers and other siblings. For example, in the Karamoja, Uganda famine in 1980/81 where there were gross food shortages affecting all age groups, a chain of centres was developed, located to maximise coverage. These were designed around a compound with shade and a single guarded entrance/ exit. The admission criteria were anthropometric and/or clinical, and any individual meeting the criteria would be admitted to receive a full cooked ration to be eaten 'on the spot'.

In Korem, Ethiopia, SC UK employed similar criteria to admit children to a large feeding centre where, again, a full ration was supplied, in this case also including carers and sometimes siblings. The most malnourished adults, from the many thousands living in the open without food, were clinically selected by MSF for admission to a relief shed.

Imposed targeting criteria and distribution methods, particularly when food aid is in short supply, can be intrinsically undignified and sometimes actively humiliating to the recipient (see Box 10). Protecting the dignity of disaster affected populations is an important human rights principle. Every effort should be made to ensure that to the extent possible, communities are consulted on the approach used and play a role in the design of the targeting system, and humanitarian agencies take action to secure further resources for the affected population where humanitarian assistance falls far short of need.

The time and expertise required to conduct community managed targeting effectively has been well documented and should not be underestimated. Where the time demands on the Relief Committee are excessive, consideration needs to be given to some form of compensation (Shoham, 1999). Once the distribution is up and running, there may be a reduced cost to the external agency as the community takes responsibility and any field presence is limited to a monitoring or supervisory role. The length of time which an agency has been in a particular area, as well as a range of other factors, will affect the amount of time required to start community managed targeting.. For example, in 2001 in Binga, Zimbabwe, Save the Children aimed to reach a total of 51,000 beneficiaries, in approximately 120 villages, with food aid targeted in the community (see also Table 4). If conducted according to the steps outlined in box 14, it would have taken 14 weeks to complete the preparatory work, employing 6 field officers (see Box 22). The original intention was to distribute in weeks 5-8 and 9-12 and yet the entire process, if completed, would have taken 14 weeks. In the end, it was decided to omit stage 5 (see Box 22) and thereby, start distribution in week 10. Targeting food aid, using the methods described in Sections 2-5 for reaching households, are likely to fail in these circumstances. While it may not be possible to avoid the approaches outlined in these sections, high levels of inclusion and exclusion can be expected.

• Box 16 describes the system of relief committees set up in southern Sudan with a strong representation by women. In 1998, there was an extreme crisis with very high mortality. The targeting system described was abandoned by the community and every distribution was subject to redistribution. The main justification given for switching was the deterioration of the situation and "the fact that the Chiefs are solely accountable and responsible for hunger related deaths amongst their communities". In addition, a tax was imposed on the distribution system to support "government authorities" (Chapman 1998b).
• In north east Amhara region, Ethiopia, in 1999, food aid provision (through free food distribution and the employment generation scheme) was insufficient to prevent widespread hunger and elevated rates of malnutrition. In woredas where the needs were greatest, local authorities redistributed the food aid to 100% of households because food aid allocations were perceived to be insufficient relative to local needs and everyone was considered affected by the crisis. In reality, the variation at local level in the actual targeting was substantial.

Community based targeting in Binga, Zimbabwe

In addition to errors, there may be other consequences of pursuing these strategies in times of acute shortage, i.e. social division and conflict. In 1996, Oxfam undertook a targeted food distribution in Turkana district in Kenya. Under pressure to limit the extent of the food distribution, the community was asked to identify only the "worst cases" (estimated by the local famine early warning system to be approximately 30% of the population). An unexpected time lag between the time of the assessment and the first distribution meant that the situation had deteriorated and, at the time of the distributions, the community felt that many more people should have been eligible for assistance. In fact, they were having to choose between the destitute. Table 5 below shows the problems highlighted by Oxfam when the programme was evaluated according to each distribution centre. Oxfam notes that it was remarkable that the community agreed to actually apply the targeting criteria given to them by Oxfam, and that this was probably due to the long and well-established relationship that the agency had with the affected community.

Cost of targeting

It is generally assumed that targeting systems which require higher degrees of accuracy require greater resources (see Table 6). However, the evidence for this is limited. Each type of targeting system carries different types of budgetary costs and human resource requirements. Some carry costs for the implementing agency, some carry costs for the community.

Children may be a group targeted to receive food, when needs exceed what food aid is available

The costs of a particular targeting system should also be considered in relation to the extent to which objectives have been effectively achieved. The cost effectiveness of general distribution, in comparison to targeted supplementary feeding, was reviewed in Hartisheik A camp in Ethiopia for Somali refugees, 1988-1989 (AbuSaleh, 1993). The results showed that a dry supplementary feeding programme, targeted to moderately malnourished children under five, and a general ration averted malnutrition at a cost of $US221 per case. If, rather than delivering separate supplementary feeding services, the general ration was improved in quality (rather than quantity) to include the increased nutrients available in the supplementary ration, the cost fell to US$118 per case averted. The study also examined the cost effectiveness of wet supplementary feeding, both in the context of a small general ration and a large general ration. The analysis found the cost per case of malnutrition treated was US$437 per case in the first situation, and $56.7 in the second situation. The authors concluded that a high quality general ration was significantly more cost effective in preventing malnutrition than a small general ration coupled with wet feeding.

Box 22: Time table of preparatory work for community based targeting in Binga, Zimbabwe