The wide variation of social contexts in which different activities are undertaken will often lead to wide differences in the final choice of targeting method.
Once a rationale for targeting has been defined for a food or nutrition programme, the selection of a targeting scheme becomes part of the programme design and development process. A programme's specific objectives are particularly relevant to targeting. The at-risk population, and thus the target group, and the programme objectives are identified through a problem analysis, a vulnerability assessment and a poverty profile which are usually the first steps in the programme design process. Within this context, it is important to make a distinction between food- or nutritionally insecure individuals or households, and individuals or households that are vulnerable to (or at risk of) food and nutritional insecurity (see the Box on Targeting terms in Chapter 1).
It should be remembered that there are no simple rules for selecting an appropriate targeting method. The wide variation of social contexts in which different activities are undertaken will often lead to wide differences in the final choice of targeting method. For example, the determinants of malnutrition vary greatly from place to place and across populations. As a result, the best indicators for predicting the risk of future malnutrition for screening purposes are also likely to vary.
In addition, selection of the best targeting strategy depends on the type of programme. For example, a programme involving on-site feeding could take advantage of the gathering of beneficiaries to carry out a continuous screening programme, such as anthropometric screening that calls for monthly weighing. Such screening in a take-home feeding programme would require the additional effort of bringing a weighing scale to each participant and, thus, would have to rely on less frequent weighing.
Generally, targeting strategies must be consistent with the existing level of infrastructure (logistical, human resources, institutional) and the area, and must be conceived within the political, financial, cultural and technical constraints of the programme.
Any intervention may be targeted by using multiple methods and multiple selection criteria. Combined methods are often used to strengthen overall targeting efficiency. Geographic targeting may be combined with community, household or individual targeting mechanisms. For example, emergency relief food is often targeted first on a geographic basis, according to the general severity of the crisis in different regions, and then through household screening criteria that rely on community-based channels to allocate relief to specific households (see the Ethiopia case study in the Annex.)
Food and nutrition programmes can be targeted through various schemes. Essentially, programmes are either administratively targeted or self-targeting. Administrative targeting involves the selection of specific regions, areas or communities (geographic targeting), or of specific households or individuals. The latter may be done by programme planners and staff, or by communities (community-based targeting). Self-targeting does not involve the selection of programme participants, but instead relies on programme incentives to induce maximum participation by members of a target group. When programme incentives are introduced through changes in market forces, they are referred to as market-based targeting.
Targeting can also be a multi-stage process. For example, certain regions may be targeted first because the food or nutrition problem is deemed to be most severe there, perhaps resulting from a natural disaster. Next, within each selected region, communities that meet certain criteria may be selected, such as communities with a high percentage of woman-headed households or communities with a high proportion of internally displaced households. After that, specific households and/or individuals in the selected communities who meet certain eligibility criteria may be targeted. The targeting mechanism is the application of a set of rules through which it is decided whom to include and whom to exclude. The different targeting mechanisms are explained in greater detail in the following sections.
In an administratively targeted scheme, decisions on the eligibility of individuals or groups (including regions) are determined by programme staff on the basis of whether or not candidates meet defined eligibility criteria. These are based on one or more indicators that have been previously defined for the purpose of targeting. Administrative targeting may also be based on a so-called "means test", i.e. in order to qualify for participation in the programme, the household or individual must not have the means, or a minimum set of assets, to obtain an adequate level of food intake (this can be decided in terms of, for example, per capita income, amount of land under cultivation or herd size). Means tests may be established on the basis of targeting indicators, but such tests are usually more costly to apply and administer. Indicator-based targeting usually involves the collection and analysis of data on those targeting indicators that are thought to be the most objective and representative of the defined selection criteria. The criteria must be defined and measured consistently for each candidate individual, household or group of individuals or households.
Most forms of targeting include an administrative element in the definition of programme areas, the definition of targeting criteria or the definition of target groups. Each form of administrative targeting may be used alone or in conjunction with other methods of targeting. For example, the selection of different markets for market-based interventions or of different communities for self-targeting public works activities represent forms of administrative targeting.
The first step in the development of an administrative targeting methodology is assessment of the various levels of administration and decision-making that are necessary to implement a programme, and of the information available on which to base decisions for targeting at each level. In many cases, programme resources will be managed initially at the central level and allocated incrementally to lower levels of administrative decision-making, until they finally reach the beneficiary population.
The application of administrative targeting schemes may become complicated in countries where there is no clearly articulated national food and nutrition policy and where different levels of administrative decision-making (e.g. national-, regional- and local-level) may lead to different sets of eligibility criteria at each region or locality. Locally determined eligibility criteria may have greater validity than uniform criteria that are nationally applied.
Once the administrative structure and decision-making needs have been defined at each level, administrative targeting requires the following additional tasks:
Information costs depend largely on the deployment of programme staff and infrastructure in relation to the location of the beneficiary population, as well as on the information collecting mechanisms used. When staff are located close to the intended target population, such as when clinic or school staff collect targeting information, data collection costs may be marginal. Similarly, when the service delivery mode requires candidate participants to present themselves at a central location to apply for benefits (e.g. an MCH centre), data collection costs may again be low, although this system can also lead to a large exclusion error in targeting if no additional efforts are made to locate eligible households or individuals. In contrast, where programme staff have limited and non-constant access to the intended target population and must incur substantial time and transportation costs to collect targeting information, administrative targeting is likely to be costly.
Community-based targeting schemes depend on decision-making structures at the community level to allocate programme goods and/or services effectively according to the criteria defined in the programme objectives. In some cases, the communities themselves may contribute to defining these objectives through a participatory decision-making process. In other cases, such as in emergency feeding activities, the objectives may be more self-evident and universally applicable. As communities can interpret criteria differently, leading to differences in the allocation of relief food supplies, it is important to ensure that community members understand clearly the programme objectives, and the basis on which relief supplies are allocated, prior to programme implementation.
Effective community-based targeting requires preliminary efforts in the following areas:
Assessment serves to examine what the likely outcomes will be from a community-based targeting scheme, in terms of who in the community will be reached by the programme and the extent to which the programme can be expected to deal effectively with food and nutrition problems in the community. The assessment may also help to determine whether the programme needs to include specific activities to strengthen the decision-making structures in the community.
The frequent and participatory auditing of community-based targeting outcomes is very important to ensure that targeting errors are identified and corrected. Such auditing can serve as a monitoring tool and can help communities to review and adjust local targeting practices.
Community-based targeting is often influenced by a few local leaders and administrators, and this can result in the subjective selection of beneficiaries that is not based on a needs assessment. In local settings, people often follow their leaders and do not have a real space in which to express their honest opinions. At the same time, programme staff cannot approach a community without directly involving its leaders and political decision-makers. Their best way of dealing with this situation is to promote and facilitate the creation of a more participatory decision-making environment, starting perhaps with a participatory needs assessment in the community.
The central aspect of self-targeting mechanisms is that the decision of individuals or households on whether or not to participate in a given programme is the main determinant of who receives programme benefits. In other words, self-targeting occurs when benefits are available to all, but programme incentives are set in such a way that the non-needy elect not to participate. While other targeted programmes require social workers or other human resources to establish eligibility criteria, screen and monitor eligibility, with self-targeting the decision to participate is made by the households or individuals themselves.
Universal access to programme benefits in the form of goods, services or employment is necessary in a purely self-targeting scheme. If programme benefits need to be rationed because of resource constraints, self-targeting may take place in combination with some form of administrative or geographic targeting. For example, if a food-for-work programme intends to generate non-farm income for women in a drought-affected region, a combination of geographic targeting, administrative targeting (only women are eligible) and self-targeting (food ration levels are set at such levels that only the neediest women are likely to decide to participate) will be involved.
Three main elements influence an individual's decision to participate in a given activity:
A key to successful self-targeting is the clear definition of the target population during the programme development stage, such that the offered benefits are likely to be demanded only by that target population at a price that only the target population is willing to pay.
Effective self-targeting of food programmes requires detailed information about market conditions with regard to supplies and prices of specific foods, and the income and the consumption patterns of different segments of the population (i.e. price elasticity of demand from various income groups). The key is to segment the population in order to identify major differences in the preferences and market behaviour of various groups, such that differences in the type, quality and cost of the food item(s) offered lead to self-selection by the intended target population. For example, when there is a reduction in the price of a food product that most consumers consider to be of inferior quality and/or low social value, the demand for that food is likely to increase substantially more for poorer groups than for higher-income groups. The use of a quality differentiation device (based on market mechanisms) to promote self-targeting schemes has been successful in many developing countries, especially in food subsidy programmes. However, a number of important considerations must be taken into account if self-targeting is to work (see the Tunisia case study in the Annex.)
The design of a self-targeting food subsidy programme that is based on quality grading involves examining household expenditure data to determine whether there are significant differences in consumption patterns across income groups, especially with regard to basic food commodities. For example, yellow maize flour is considered to be inferior by consumers in Central America compared with white maize flour. Yellow maize flour (which is of higher nutritional value) is consumed by the poor, and white maize flour by higher-income groups. Where the poor consume a different basket of goods from higher-income consumers, the price of one or all of the foods contained in that basket is to be subsidized. Quality grading can also be introduced through different packaging, colour coding and labelling, which provide the image of a superior food by giving it more attractive packaging without basically changing the food commodity itself.
An important issue involved in self-targeting food programmes is the question of consumer acceptance. Care must be taken that, by making a food product unattractive to higher-income consumers, it does not also become unappealing to the poor.
A good example of the importance of consumer acceptance occurred in Tunisia, where the introduction of a heavily-subsidized and less refined brown sugar was unsuccessful. Despite the fact that the new subsidized sugar was almost half the price of commonly used white sugar, it was not readily accepted by consumers at any income level, apparently because darker sugar was assumed to be "dirty". The failure of this intervention was mainly associated with programme designers' lack of prior understanding of consumer acceptance, and the absence of accompanying efforts to raise consumer awareness and provide nutrition education.
Some examples of activities with built-in self-targeting mechanisms are:
The self-targeting that results in the last example is (or should be) unintentional, in the sense that efforts should always be geared towards improving public health and education services.
Food subsidies are one way of increasing the purchasing power of the poor and compensate for losses in real income caused by economic crisis, unemployment, income gaps or disparities, or wars. Subsidies serve as part of social safety-net measures. In several countries, such programmes have had a positive effect on household food security and nutrition, especially among urban populations. It has been reported in India and the Philippines that these programmes have reduced the prevalence of underweight children. Some food subsidy programmes based on foreign food aid have helped to offset the fiscal costs of food subsidies, and contributed to increased public investment in social services. Current economic crises are forcing many governments to look into the use of self-targeting for subsidy programmes as a means of reducing budgetary costs and finding more efficient ways of directing benefits to those most in need.
In practice, the development of self-targeting activities may also include dimensions of geographic targeting, by locating public works programmes in specific geographical areas or distributing subsidized food commodities through retail outlets in low-income neighbourhoods.
Market-based interventions aim to increase the demand or supply of a particular food among the target population.
Market-based targeting is very similar to self-targeting, as it depends on the choices of individuals to buy or sell goods or services in the market. The basis of market-based targeting is the supply-demand equation. The application of both market-based and self-targeting methods relies on the assumptions that:
Market interventions may be oriented to influence market supplies of foods and services available for purchase; for example, through the local sale of grain from reserves or the marketing of oral rehydration salts or locally produced weaning foods. These types of interventions may also address market demand, either directly through official market purchases of commodities to support local prices and the incomes of producers, or more indirectly through public education and social marketing intended to influence consumer preferences for key goods and services.
Market-based interventions may also choose to reach their objectives through price means, with the aim of increasing the demand or supply of a particular food or service among the target population. This can be done through price controls or price subsidies of key food commodities or services so that consumers choose to purchase more of a particular food or service. Nowadays, price controls are applied less and less frequently, while price subsidies (which are also less frequently applied in market-oriented economies), when applied to basic foods, closely resemble self-targeting mechanisms. Improving the productivity of domestic food producers and expanding commercial food imports are more sustainable ways of increasing domestic market supplies of food and decreasing domestic food prices.
A clear understanding of the different characteristics of a selected market, including its level of integration, is essential in determining the appropriate intervention, its probable effectiveness and coverage and the level of resources necessary to meet the stated objectives. For example, decisions on whether to employ cash-for-work or food-for-work as a means of transferring purchasing power to poor households depend critically on the level of market integration in the programme area. When markets are well integrated and food supplies flow freely in response to changes in demand, the use of cash transfers is likely to minimize the distorting effects of food transfers on local market incentives. Similarly, when markets are well integrated, the price effects of a local food purchase activity may be diluted over a broader area, requiring larger purchases to bring about a given price change.
Geographic or regional targeting can be accomplished through a variety of methods. The simplest form is based directly on local estimates of need for demand-driven activities, where resources are allocated to various regions in proportion to stated levels of need. However, local needs are likely to be overestimated, as part of the inter-regional bargaining process for programme resources. A more complicated but increasingly common form of regional targeting involves poverty, food insecurity and vulnerability assessment and mapping systems, which use rapid rural appraisal, participatory rural appraisal or rapid food and livelihood security assessment methodologies. The construction of a poverty map based on a composite poverty index is another method (see the geographic targeting element of the national school feeding programme in the Costa Rica case study in the Annex). Such efforts utilize a range of information sources and indicators. They usually involve the development of a multivariate index or statistical model aimed at capturing the basic dimensions of the stated targeting criteria by using different socio-economic indicators such as literacy rate, female education rate, unemployment rate, level of income, agricultural failure, population growth rate and prevalence of malnutrition.
Early warning information systems contribute information and data on emerging food crises in specific locations, based on agroclimatic conditions, food crop estimates and food market conditions as indicated by local food prices. Regional food balance estimates, particularly when they are calculated on a monthly basis, may also provide early warning information regarding food shortages. Structural vulnerability assessments and detailed food and livelihood insecurity analyses provide information and data for the design and implementation of local development programmes.
In some instances, targeting broadly defined geographic areas has been found to offer only minimal gains in programme efficiency compared with cases where no targeting is used. Studies using household survey data indicate that the use of smaller geographic units in geographic targeting can reduce programme leakage and improve the coverage of the target population as the unit targeted becomes smaller.
Although regional targeting is sometimes carried out, the methodologies for doing so efficiently and effectively are not always well developed (see some of the country case studies in the Annex). For example, it is often assumed that regional disparities in food security and other welfare measures can readily be understood and identified through the informed judgement of local informants and decision-makers. However, biases in the perception of the determinants of food insecurity have led to serious inefficiencies in geographic targeting. To avoid such biases, there is a clear need for a more focused understanding of regional food security and nutrition conditions. This is particularly true when emergency food relief is being allocated among the various regions affected by a natural or human-incurred disaster (see the Ethiopia case study in the Annex).
In fact, nutrition programmes may be geographically targeted, but the selection of service area boundaries may not always be based on nutritional criteria. Non-nutritional concerns frequently dominate when broad service areas (regions, provinces, states, etc.) are designated, and nutritional concerns may govern the selection of specific villages, households and individuals within the broadly defined service area.
In Burkina Faso, food aid that was targeted on the basis of the regions' different agricultural potentials resulted in the distribution of large amounts of food aid resources to relatively better-off households. In fact, because households in areas of low agricultural potential had developed diversified income strategies, their access to food was less severely affected by adverse agroclimatic conditions than that of households in areas of higher agricultural potential which were more dependent on agriculture as their main source of income and food.
In most circumstances, small area estimation techniques have been shown to produce statistically reliable estimates of variables for small geographic units. While already used frequently in the United States and other developed countries, small area estimation methods are now being applied with greater frequency in developing countries as well. By providing statistically rigorous estimates of targeting indicators at the local level, these methods can substantially improve the efficiency gains from regional stargeting.
Geographic targeting for drought relief in Ethiopia
In Ethiopia, a map showing chronically vulnerable and food-insecure areas has been constructed on the basis of a composite index of nine indicators. The map rates weredas (districts) as "very highly vulnerable", "highly vulnerable", "moderately vulnerable", " slightly vulnerable" and "very slightly vulnerable". No data are available for two departments (Afar and Somali). The indicators used as inputs to construct the index are:
Each of the nine indicators was weighted, first individually by a number of collaborating agencies (Drought Prevention and Preparedness Commission [Early Warning Unit], Ministry of Agriculture, CIDA, SCF UK, the United States Agency for International Development [USAID], the European Community [EC] and the Vulnerability Analysis and Mapping Unit of the World Food Programme [VAM/WFP]) and then by applying an average group weight to each indicator. The index is currently being validated in the field in selected weredas.
The selection of target groups, households or individuals is often determined through a multistage targeting approach: first, an administrative and/or a geographic targeting approach leads to the selection of broad service areas (regions, groups of villages, provinces, states), then household- or individual-level eligibility criteria are applied. Target groups are selected through the use of specific criteria that discriminate the most needy from the less needy within a given community.
Choosing the most appropriate criteria is the key to minimizing inclusion and exclusion errors when potential beneficiaries at the household or individual level are being screened. Such criteria are based on the characteristics of households or individuals.
Infants and young children are particularly at-risk of not getting the food they need for good health, growth and development.
For nutrition programmes, households are often screened mainly on the basis of their size, their socio-economic status, the education level of the mother, child spacing within the family, and/or a history of poor nutrition for any family member. A household's socio-economic status can be a useful way of identifying malnourished individuals of those at high risk of malnourishment. Relevant socio-economic indicators include individuals living in families or households in which an infant has died, in which the mother is very young or relatively old, or in which there are many young children.
Targeting all the at-risk individuals in a nutrition programme area is basically done through such indicators as:
A cross-country study demonstrated that a number of simple indicators (each measured by only two or three different values) performed quite well, singly or in combination, in identifying households and children under five years of age at risk of food and nutritional insecurity. These indicators include: number of individual foods consumed; household size; household dependency ratio; number of rooms per person; incidence of illness; vaccination status; age at weaning of children under five years of age; and drinking-water and sanitation facilities.
Any of these indicators can be applied, singly or in combination, when selecting potential recipients. The screening indicators most commonly used in individual targeting are age and anthropometric measurements. The concept of selecting individuals of a specific age cohort has been developed because certain age groups are at risk, or high-risk, of developing a state of malnutrition. Examples of such age groups and the particular risks that they face are osteoporosis in elderly women and malnutrition in small children of 6 to 36 months of age.
The most useful anthropometric indicators are:
Weight-for-age and height-for-age are the most commonly used measures of malnutrition in children under five years of age, although in emergency situations weight-for-height or MUAC are often used to identify the most at-risk children. When severe protein malnutrition is a main problem, as when kwashiorkor develops, weight-for-age may provide a false assessment because of the presence of oedema, and thus there is the possibility of an exclusion error. The rationale for using weight gain criteria for targeting is that a child under five years of age who fails to gain weight over a period of time is at extreme risk of malnutrition. In areas where malnutrition is prevalent, weight gain criteria screen out, relatively effectively, the children aged under five years who have a normal nutritional status but miss a large proportion of those whose nutritional status would deteriorate over time in the absence of supplementary feeding. A combination of age and anthropometric values is also often applied to target more narrowly the children who are most at risk, say those in the 6 to 36 months age group.
Generally, nutritional criteria may be appropriate for targeting individuals, while socio-economic factors may prove more useful when selecting households for targeting.
It is important not to exclude a particular programme objective or targeting criterion simply because it seems to have low cost-effectiveness. While screening each individual according to the given criteria may seem costly in the short term, the programme will benefit from such screening in the long term because it minimizes exclusion and inclusion errors.
The use of anthropometric indicators in targeted group feeding programmes (GFPs)
ANTHROPOMETRIC INDICATORS ARE USED TO:
SELECTED INDICATORS IN EMERGENCY SUPPLEMENTARY FEEDING PROGRAMMES
Apply weight-for-height (or weight-for-length) in children. The priority is to select wasted children because they are at most risk of serious illness and death. If there are many children, a preliminary screening using mid-upper arm circumference (MUAC) can be used. The cut-off levels used to select children vary according to the resources available, but one recommendation (ACC/SCN, 1990b and UNICEF, 1986) is:
Children whose growth charts show poor weight gain can also be selected.
SELECTED INDICATORS IN NON-EMERGENCY SUPPLEMENTARY FEEDING PROGRAMMES
Since there are likely to be fewer wasted children in non-emergency than in emergency situations, if resources permit, some stunted children can also be selected. The following indicators and cut-off levels are recommended:
If height cannot be measured, use weight-for-age of less than 80 percent or, for one- to five-year-olds only, MUAC of less than 13.5 cm.