Special Focus on Targeting

by Jeremy Shoham

X marks the spot for the food air-drops in South Sudan (WFP 1998, F. Pagetti)

The recent spate of reports and articles on targeting emergency food aid shows the resurgence of interest in this subject. This is not however a new subject. During the Sahelian famine of 1984-5 many international aid agencies put considerable effort into collecting information to help target food aid across the region. The reasons given for targeting at that time were the same as current justifications, i.e. the desire to:

• reach only those in need of food aid or sometimes those most in need given scarce food aid or logistic resources;
• avoid creating dependence by over-supplying food aid;
• limit disruption to the local economy through oversupply, e.g. through an effect on market prices.

In some emergency situations targeting (which involves excluding some of the population from food aid assistance) will not be attempted as it will be obvious that the whole population is in need (e.g. Somalia 1992-3). In these cases the emphasis will be on how to reach all with necessary food aid so that different food distribution mechanisms may be employed e.g. general distributions, soup kitchens, supplementary feeding.

One of the main findings of a review of the targeting experiences of the 1980s was that the types of information used to inform targeting decisions had changed from previous decades. Previously targeted programmes had mainly relied on data on agricultural production and impact of food deficits on nutritional status. More recently, needs assessments and food information systems have increasingly used data on population/household access to food, income sources and types of survival strategy employed by the affected populations to inform targeting decisions. This shift in methodology appears largely to reflect the gradual adoption of Sen's model of famine, e.g. that famine results from failure of entitlement rather than of food supply.

Targeting once again appears to be a major preoccupation of aid agencies and governments. In the famine in southern Sudan last year much effort was made to design emergency feeding programmes which would target the neediest. The experiences triggered a number of reports and articles on the subject (see Field Exchange Feb. '99 ). For the purposes of this special focus article we will distinguish three stages of targeting:

• at the national level (stage 1) governments, donors and national early warning systems are involved in decisions to identify and prioritise areas in need of food aid (often called geographic targeting) and in the design of appropriate interventions, At this level information systems are central to targeting decisions
• at the sub-national level (stage 2) there is a further refinement of geographic targeting in which local government and implementing agencies decide on the distribution within the region or district to community or village level. This is also usually based on some form of information system.
• at the beneficiary level (stage 3) household or individual recipients are selected by administrative systems, e.g. through village committees or household surveys or through the mechanism of distribution, i.e. the choice of mechanism such as food for work or school feeding determines who gets the food.

Targeting cannot be attempted without information on food security, i.e. food and income sources and coping capacity, at the regional, sub-regional or household level.

Stage 1 and 2 targeting are rarely easy. In some situations there are no existing food information systems or assessments with baseline data against which to compare current information. Furthermore, certain types of information are difficult to quantify or interpret, e.g. livestock status or collection of wild foods. Value judgements may also need to be made, e.g. should charcoal production or sale of livestock assets be counted or discounted as resources. If the view is that charcoal production denudes the environment while sale of key livestock assets undermines long-term viability, then food or income derived from these resources may not be included in assessing the degree of self-sufficiency of a population or group. This in turn affects the assessment of overall food needs of a community/population. Stage 2 targeting can also be extremely problematic from the point of view of local political pressures, especially where programmes are implemented by local government rather than outside agencies. Nevertheless, in spite of these difficulties there seems to be an increasing ground-swell of opinion that the greatest gains in targeting efficiency can be made by improving information systems which inform decisions about geographic targeting rather than by refining intra-community targeting systems (stage 3).

The jury on intra-community targeting (stage 3) is still out. In the famines of the early 1980s experiences were mixed. An Oxfam evaluation report from the Red Sea Province in 1984 concluded that efforts to target by household were thwarted by the Beja nomad view that food aid resources should be shared equally amongst all households irrespective of socio-economic status. This contrasted with experiences in South White Nile province in northern Sudan where household level targeting appeared to be accepted by the community. More recently, efforts to establish distribution committees in southern Sudan which identified and targeted the most vulnerable households were largely unsuccessful as once monitoring agencies had left the distribution site, a redistribution would take place with food often being shared rather than targeted. In contrast, village committees established in Central Tanzania during 1998/9 to distribute food aid to drought affected households were apparently successful in identifying and targeting the most needy households. In this instance the committees in conjunction with the community established their own eligibility criteria. It could be argued that certain identifiable factors determine the success of one programme and the failure of the other (see Community Managed Targeting of Emergency Food Aid: Does it Ever Work?, Field Exchange 7, July 99). One factor which can 'de-rail' outsider attempts to implement intracommunity targeting is that outsider perceptions of vulnerability are often substantially different to those of local communities. Local political structure and dynamics are also key factors in determining the feasibility of intra-community targeting and should form a critical component of any needs assessment. It has recently been argued by some observers that in almost all situations of food crisis, fair and needs based intra-community targeting is effectively precluded by political processes within communities.

Sudan

Another factor which may account for the variation in successes of intra community targeting, is the severity of the crisis affecting the community. There may be a point of severity beyond which targeting will not work in any community.

While the mechanism of food distribution can support certain forms of targeting, there are difficulties and uncertainties connected with each type of mechanism which make it difficult to recommend one type of intervention over another.

For example, one rationale for Food For Work programmes (FFW) is that such programmes selfselect the neediest members of a community. However, recent findings in Ethiopia calls this into question as it appears that households with most labour to spare may willingly participate in the work. Cash for work (CFW) is also being increasingly advocated amongst some parties, especially as this form of intervention would be cheaper since bulk food aid does not have to be moved long distances and it would stimulate/encourage local/regional markets. However, there has been very little empirical study on CFW projects and it is possible that these programmes 'self-target' even fewer needy households than FFW. Also, CFW may prove more gender divisive than FFW while in situations of conflict the injection of cash may increase insecurity.

Distributing food aid through existing infrastructure, e.g. school feeding or MCH clinics, is a means to reach a demographic target group, but these may not necessarily be the most needy, e.g. only the better off may go to school or be near enough to attend health centres. Furthermore, targeting food aid to households on the basis of a child's nutritional status (as often occurs through MCH programmes) may target households with care or health problems that result in malnutrition, rather than the most food insecure households.

Market interventions, i.e. putting food aid on to the market in order to reduce prices and increase availability, would also on the face of it appear to be a cost-effective self-targeted intervention. In theory, the quantity of food that would need to be released onto a market to substantially reduce prices would be far less than the quantity of food needed to feed a large population through a general ration programme. However, the experience of managing market interventions in emergencies has been extremely mixed and have shown that this type of intervention requires great expertise and commitment from government.

It seems obvious that if it is thought appropriate or desirable to target food aid in a given context, the feasibility and means of such a strategy should be assessed. This would have implications for the design and methodology of food security assessments.

In most situations it should be possible to classify the circumstances or factors where certain forms of targeting are appropriate and feasible. This is, however, a domain where we lack guidelines or good practice reviews.

Looking back at the past 15 years, it appears that there has been little headway in defining when and how to target emergency food aid. Yet there has been substantial documented experience of attempts to target food aid resources. Perhaps we are now in a position to begin the process of identifying those circumstances which indicate a predisposition towards one form of targeting rather than another and to move towards producing guidelines for field staff faced with this most difficult and responsible programming decision.