Determining eligibility (Special Supplement 1)
Children are often a group targeted in emergencies
Eligibility criteria, i.e. the characteristics of those individuals or households to be targeted with food, arise from the objectives. Clearly, if an objective is to meet the needs of a group of individuals or households which are thought to require a certain quantity and quality of food, then the eligibility criteria must specify the characteristics of these individuals or households. Criteria need to be both sensitive (to ensure that those eligible are not excluded) and specific (to ensure that those not eligible are excluded).
There are two aspects to determining eligibility. The first is the defining of the eligibility criteria and the second is applying them in practice.
The eligibility criteria adopted must ensure a practical way of identifying individuals or households at the point where they receive food. If, for example, the eligibility criteria set for eligible households are that the household must have less than a particular amount of land or livestock, there must be some method of checking that the household does, in fact, have this characteristic. A targeting system can be no more accurate than its ability to identify the beneficiaries. Eligibility criteria must therefore be clear (see Box 5).
Eligibility criteria fall into two broad categories:
Criteria applied to individuals. The need to target individuals in emergencies usually arises because inadequate support has been given to households, e.g. because of early warning failure, shortage of relief food, or because they have been excluded in some way from support to the general population. Individuals are targeted either because they are currently malnourished, e.g. children less than 80% of the median weight for height, or because it is thought that they have special needs and are unlikely to be able to meet their food needs, e.g. the sick, particular ethnic groups, pregnant or lactating women, the elderly, the disabled, or orphans.
Box 5: Unclear targeting criteria
In Tanzania, drought relief in 1998 registered people in affected areas in
- people with no means to buy food and unable to work
- people with no means to buy food but able to work
- people with the means to buy food
Group 1 was to receive free food, group 2 food for work, and group 3 no
food aid or, perhaps, the opportunity to buy food from government stocks.
In one area, category 3 was taken by local leaders to mean that part of
the food aid allocation could be sold to raise funds for community work,
such as school repairs. In practice, this had (not surprisingly) led to the
better-off benefiting more than the poor from highly subsidised food sales,
and to suspicions and conflicts about the actual use of the proceeds.
Criteria applied to households or groups of households. The intention is usually to make up for some measured or assumed deficit in the household's ability to acquire food. The criteria applied are usually socio-economic, e.g. households with less than a certain amount of land or livestock, or households that are displaced. The criteria are usually proxy indicators which are associated with the target group. Criteria may be defined by outsiders or by the community themselves. However, a common understanding of the definition of the household must be reached between the recipient population and the external agency. A recent evaluation of community managed targeting in Malawi showed that households tend to reside in clusters which are economically interdependent. In this situation, targeting by household resulted in considerable redistribution (see Section 5) (Mathys, 2003). In pastoralist communities, households may be routinely split up as different members pursue distinct, but complementary, economic activities.
No eligibility criteria. Households and individuals can also be targeted indirectly through self-targeting systems.
Each of these categories is discussed in detail below
Targeting individuals Eligibility criteria applied to individuals vary in the ease and accuracy with which they can be applied. For the most part, criteria applied to individuals are set by outsiders to the community (also called administrative targeting).
Targeting individuals according to nutritional status
Anthropometric status provides an objective basis for the selection of individuals to receive food, but as noted (Section 1), can exclude other needy individuals. The anthropometric, and other, criteria for admission for supplementary or therapeutic feeding are largely standardised and widely agreed (WHO, 2000). Criteria may be adjusted in situations of extreme crisis or where resources are inadequate to deal with the caseload of malnutrition, e.g. the weight for height required at discharge may be reduced to allow more children to pass through the programme, and/or the entry criteria may be lowered so that care is only directed at the most malnourished (for an example, see Maxwell, 2000).
Box 6: Emergency school feeding
In Kakuma refugee camp in Kenya, school feeding was
established for adolescent boys. This decision was based
on the experience of Sudanese refugee camps in
Ethiopia, where boys were often living on their own and
where the general ration to which they were entitled
(calculated on the basis of an average population
requirement) was insufficient. While planners in Ethiopia
were aware of calorific shortfall in the ration received by
these young men, an assumption was made that they
had opportunities to earn additional income and could,
therefore, meet the outstanding deficit. In fact, this was
not true and as a result, many became acutely
malnourished. School feeding for adolescent boys was set
up in Kakuma to avoid this situation from recurring.
Nutritional clinical criteria may also be used, e.g. nutritional oedema, people in obviously poor nutritional condition, or those with overt vitamin deficiencies can be admitted to feeding centres (Collins et al, 2000). Such criteria are, however, difficult to standardise so that it may be difficult to apply these where admission is controlled by inexperienced staff.
Targeting individuals according to health status
A school feeding programme in Zimbabwe
Targeting individuals with food according to their disease status is virtually impossible unless a) there is an existing, accurate means of identifying the recipients (such as a TB clinic or hospital), or b) very broad definitions are used, which could result in high levels of inclusion error.
Targeting pregnant and breastfeeding women
Pregnant and lactating women are relatively easy to identify. This group is often targeted with food - usually through dry supplementary feeding because it is easy to do and can be used to provide a mechanism for referral for antenatal care, micronutrient supplementation or immunisation. The actual vulnerability of pregnant and lactating women in terms of food shortage will depend on whether the general ration is adequate (see Section 1). In a situation where the general ration is adequate, but it is believed that women are not receiving sufficient food because of biases in intra-household allocation, there would also be a logical basis for targeting this group with food.
Targeting the disabled and elderly
The elderly may be nutritionally vulnerable. Reduced physical or mental function may make it difficult for them to access food, particularly in situations of displacement where social support networks or access to traditional foods is disrupted. The nutritional vulnerability of older people should not be assumed in every context, but older people may be nutritionally vulnerable in a situation where the majority of the population are older people (if, for example, the remainder of the population has fled or migrated) (HelpAge International, 2001). The definition of the elderly in a general population may be difficult3. Equally, trying to define disability is a complex and controversial matter. Acceptable terminology changes over time, and from one culture to another (Harris and Enfield, 2003).
Targeting school children and people attending or resident in institutions
School enrolment varies enormously from country to country but, in parts of some countries, can be as low as 10%. Generally, children who are enrolled in school are likely to be from higher and more powerful socio-economic groups, better nourished, and are more likely to be boys. Therefore, they are, in general, less likely to be vulnerable to food and nutrition crises than children out of school. The reverse can be true in pastoral situations, where children from richer households are away with the livestock and children of poorer households remain in towns and villages. One of the primary motives for school feeding in emergencies is to prevent children from dropping out of school. Additionally, it has been used to provide a household ration for orphans in schools, e.g. recently in southern Africa. In general, targeting school children is not a primary means of targeting food according to need in emergencies. There are occasional exceptions, where school children are targeted because they are found to have the greatest need for food aid (Box 6).
Institutions may be targeted to reach specific groups who are thought to be vulnerable, e.g. hospital patients or orphanages. These groups may face special problems in a crisis, as inmates' relatives may find it difficult to provide support, and Government support for institutions may collapse.
Eligibility criteria for households are usually 'socio-economic' and based on either:
- an estimate, based on a formal assessment, that specific types of household in the population cannot meet their survival or livelihood needs e.g. poor, female headed, HIV affected households, or
- an assumption that this is so. The assumption is usually based on an observed high prevalence of malnutrition, and is used to justify a general ration, or to target rations, to particular households. A common targeting system is to assume that households with a malnourished child registered in a feeding centre are food insecure and to provide them with a ration.
Targeting households according to socio-economic status
Women dividing Famix for distribution
In setting eligibility criteria for general food distribution to households, a distinction is sometimes made between administrative targeting, where the eligibility criteria are set by Government, donors or external organisations, and community managed targeting where the community plays an active role in defining the criteria for eligibility. In practice, this distinction is far from clear and communities are rarely left to entirely define who should receive the distributed food and how much they should receive. Strictures are placed on the community's decision making power, e.g. only 60% of households can receive food, or the types of household which should be included is predefined (see Box 7).
Box 7: Example eligibility criteria used as a guide
for communities in Binga, Zimbabwe
for refining at community level
- Size of land holdings (typically <3 acres)
- Number of cattle owned (ranged from <2 to <4, in order
to enable those with cattle to keep draught animals and
a small number of milking cows)
- No family member in permanent salaried employment,
either locally or in nearby towns
Economic assessments may lead to criteria that can be difficult to apply directly, i.e. it may be difficult to identify those households who are economically eligible to receive food. For example, during an assessment, a 'poor' population group might be defined as one which has less than a certain amount of land or livestock, criteria which it may be difficult or impossible to verify during a distribution. To use these criteria for the inclusion of an individual household in distribution, an outside agency would require a survey of all households. This may be practical on a small scale or with a densely settled population, (e.g. by using community workers who can visit each house in a camp) but is usually impractical (See Box 8). A situation where an existing administrative system, e.g. an ID registration system, provides the necessary information for accurate beneficiary identification is rare, although it is occasionally found, e.g. the systems used to determine ration beneficiaries in the 'oil-forfood' programme in parts of Iraq. In most cases involving a large, dispersed population, there is usually no practical alternative to the involvement of the community or its administration (e.g. village chief) in the identification of the beneficiaries, as they are the only people with easy access to the necessary information. This is typically done via the existing socio-political structures, e.g. by requiring village representatives to produce lists of eligible households, usually according to criteria set by Government, donors or agencies. The criteria used may arise from an assessment, e.g. households with less than a certain amount of land. Sensitive and specific criteria may be difficult to develop, as there may be multiple characteristics that define the eligible group.
Box 8: Identifying households according to
In Cote D'Ivoire in 1996/7, a food aid agency established
that traditional vulnerability criteria did not identify
households at most risk of food insecurity. The agency
identified vulnerability criteria on the basis of an
assessment, and then conducted a household survey to
establish which households met the criteria. Very quickly,
people understood the purpose of the survey and adapted
their responses accordingly.
In community managed systems, the community is used to identify beneficiaries - thus, those who are identifying the most needy are those who have the greatest knowledge about socioeconomic factors in the targeted communities. Community managed systems are usually distinguished from 'administrative' systems by the more active participation of the recipient population, rather than only its representatives, with the aim of reaching mutually agreed and acceptable eligibility criteria. Thus, eligibility criteria tend to be more subjective, complex and locally specific (Sharp, 2001). In addition, it may be easier for communities to make judgements of relative need, whereby it is possible to identify the poorest 30% of the community, rather than setting the criteria and then seeing what proportion of the population meet them. This may make food aid planning easier.
Rice bowls in rural market in Benin
The extent to which there is a conflict between donor and recipient views varies enormously from place to place, and the problem should not be overstated. Nevertheless, food will be diverted (from the donor perspective) to ineligible people, and the reverse. The extent to which the two views can be accommodated will often depend on making a reasonable 'deal' with the recipients, such as supplying a larger quantity of food than is indicated by a strictly assessed need.
In large food distribution systems, the levels of supervision may be low and the scope for inclusion and exclusion errors may be large. In any large system, communities (and, it should be said, sometimes government administrations) have considerable power to subvert donor imposed systems. This can lead, at best, to redistribution of the food (and therefore large inclusion errors) and, at worst, to a refusal by the community to participate in the programme. In practice, attempts to target precisely a beneficiary group, using imposed externally defined criteria, invites conflict with the community who may have a different concept of need and entitlement (see Section 4).
Targeting households according to illness
Families of malnourished children may be targeted with a family ration
The high prevalence of HIV/AIDS has brought the targeting of HIV affected households into prominence. HIV/AIDS can create additional risk of food insecurity and through the course of the crisis in southern Africa, attention has been given to how food insecure households, with the additional burden of HIV/AIDS, can be effectively targeted.
First, it should be noted that the impact of HIV/AIDS on household food security is still poorly understood, certainly highly variable (i.e. HIV/AIDS affects both poor and wealthy households) and depends on a range of factors, such as which household members are affected, the household livelihood strategy and the demographic profile of the household. It is, probably, a fair generalisation to say, that in most settings, the effect of HIV/AIDS (through the loss of productive household members, increased costs, e.g. from increased numbers of dependent children) is to increase the level of poverty, but that much poverty will arise from other, unrelated causes. Thus, the most food insecure households may or may not be HIV/AIDS affected. Therefore, attempts to target HIV/AIDS affected households alone are unlikely to be appropriate, as they will exclude other food insecure households.
A carer visting a home based care client in the home based care programme (see box 9)
Further, in most contexts, it is difficult to develop criteria which identify HIV/AIDS affected households accurately. Many people will not know their HIV status and even if they did, open discussion may increase the stigmatisation they face. Proxy indicators of HIV/AIDS4 may go some way to identifying these households, but run the risk of supporting households which are not food insecure, and of excluding households which require food security support.
Where the objective of targeting is to address food insecurity, it may be better to adopt established criteria, e.g. to target the poorest households and, where administrative systems allow this, to adjust this to meet the special needs of particular households, e.g. to provide a better quality of diet to households with chronically ill members (Kadiyala and Gillespie, 2003). In southern Africa in 2002-3, the approach taken was to increase the size of the general ration, on the assumption that the increased provision would reach those affected. In addition, recommendations were made to increase the quality of food provided and where possible, provide milled, rather than whole, grain (to avoid the need for HIV/AIDS affected individuals from having to pound grain). In addition to improving the quality of rations, distribution of rations need to take into account that people living with HIV/AIDS may be less mobile and less able to carry food long distances from distribution points (SADC, 2003). They are also less likely to be able to participate in heavy labour in food for work schemes (Kadiyala and Gillespie, 2003).
During an emergency, there may be scope to target HIV/AIDS affected households with food aid on a local scale. This is likely to be possible only where a long term programme for support of these households is already in place. For example, the Zimbabwe Red Cross added food aid to the package of services offered by volunteers to HIV/AIDS affected households (Khogali, 2003). The other services included hygiene training for infection management, promoting key health and nutrition messages, and working to reduce stigma. The criteria used to identify these households, 'home based care clients' and their families are shown in Box 9.
Box 9: Criteria used by the Zimbabwe Red Cross
to identify households to receive home
based care, including food aid and in the
absence of HIV testing
- Clear evidence (medical card) of a combination of
recurring infections associated with HIV/AIDS
- constant diarrhoea
- persistent coughing/recurring TB
- swollen lymph nodes
- Kaposi sarcoma
- "permed hair"
- People undergoing TB treatment
Targeting households according to the nutritional status of children: the 'family ration'
In the absence of a clear understanding of who is food insecure, some targeting strategies have been based on the assumption that if a household has a malnourished child registered in a feeding centre, then this is an indicator of food insecurity. Households with children that are malnourished are, therefore, targeted for a general household ration. However, this assumption may not be valid where a child is malnourished primarily due to other factors, e.g. disease or inadequate care.
This targeting strategy, sometimes referred to as providing a 'family ration', has been adopted in a number of emergency contexts where support for the general population is absent or inadequate (Jaspars and Young, 1995) (Box 10). While the provision of a family ration may target food to a proportion of needy households, the strategy in situations of overall scarcity has potential weaknesses;
- It may exclude households in need of food which do not have an eligible child and lead to families attempting to admit children who do not meet the eligibility criteria, causing practical problems of numbers.
- Some children may be kept in an undernourished condition to ensure that the family has access to a ration.
- Providing the ration at the point of the malnourished child's discharge (which some programmes have done) is contrary to the logic of the distribution, i.e. it keeps a household in want for a potentially long period before they receive food support.
- Where several agencies are providing services in the same area, e.g. an ame child at several centres to receive multiple rations.
In situations of less extreme shortage, targeting households according to the nutritional status of their members may be an appropriate strategy. In Afghanistan, Concern chose anthropometric indicators to target households with a comprehensive general ration of rice, wheat, oil and beans, for all household members for five months. MUAC measurements of women and children were used to determine eligibility and therefore, many children and women who were at nutritional risk, but who were not yet meeting weight for height/BMI criteria for acute malnutrition, were included. The targeting strategy allowed women to be targeted who were known to be socially and politically vulnerable (Kopplow, 2003).
Box 10: Two case studies of the 'family ration'
Following successive droughts in Mandera, Kenya between 1994-6,
in May 1996 32.4% acute malnutrition was reported by MSF. No
request for emergency food aid was made until February 1997.
MSF, serving a population of approximately 38,000 in Central
Mandera, recognised that there was a need to target food
resources, as some households were better off than others. MSF
targeted a half ration to households with at least one member in
a feeding centre, with the objective of reaching the most
vulnerable households. The consequences were:
- a substantial increase in the numbers of children admitted in
the centres - this caused a deterioration in the quality of
care which could be offered to each child, leading to
increased rates of defaulting. In addition, increased
admissions heightened the risks of cross infection.
- a substantial increase in the pressure experienced by
feeding centres staff to admit children who did not meet
the entry criteria which, ultimately, led to inclusion errors.
- high levels of readmissions and double registration in the
programme because the ration provided a resource for the
whole family. There was some evidence that children were
intentionally underfed to ensure access to food.
The strategy soon had to be abandoned as the numbers escalated out of control.
(Duce-Marques, 1998 and Vazquez-Garcia, 1999)
In Huambo, Angola, a siege town dependent on food aid, in early 2001, a programme of family rations was initiated by USAID/WFP, in place of a household ration. Agencies complied on the basis that it was the only option offered for bringing food into the besieged city, but voiced concerns that 1) the programme was unethical, in providing food only when malnutrition had occurred rather than intervening to prevent it, and 2) food would be withheld from children or children would be stolen in order for households to meet the eligibility criteria (ADACRU/ Molisv et al, 2001). The numbers of children in the supplementary feeding programme rose and active case finding had to cease because the increased numbers of children could not be accommodated.
(Dianne Stevens, personal communication).
Targeting households according to gender
Targeting households headed by females, on the basis that such households are most vulnerable to food insecurity, is another strategy that has been used by agencies However, as with all possible target groups, no assumptions should be made which are not validated by assessment about the food security of female headed households.
Supplementary feeding programme for children
In Ethiopia, a recent study found that female-headed households were four times more likely to be destitute than male-headed households (Sharp et al, 2003). If, however, femaleheaded households were used as a criterion for targeting of food assistance to the most needy, there would be a high inclusion error (as two-thirds of these households are not destitute) and high exclusion errors (because of the higher frequency of male headed households). In other country contexts, female headed households may not face economic disadvantage compared to male headed households.
Self-targeting of households and individuals
Eligibility criteria can also be set indirectly. 'Self-targeting' approaches include:
- Market interventions which allow the individual or household to choose if they acquire food and how much they acquire. Interventions include price support for staple foods, or cash distribution to increase people's ability to purchase food.
- Commodity choice can occasionally be used to increase the chances of food reaching a particular population group.
- Food for work, which is intended to exclude non-eligible people by posing a deterrent, is typically intended to 'self-select' only those people who have no more satisfactory way of obtaining food.
Market interventions may seek to:
- affect the market supply of food in order to lower food prices
- place price ceilings on certain commodities
- increase people's effective demand for food, e.g. purchasing livestock at higher than market price or distributing cash
In Ethiopia, female headed households have been found
to be more vulnerable to food insecurity
The practical challenge is to design the strategy in such a way as to maximise its impact on the most needy. Cash distribution can be targeted using similar criteria to food distribution, although security issues may preclude its use on a large scale. Food subsidies may benefit the needy group, but unless some other deterrent is introduced, will also benefit the non-needy population. The deterrents used include subsidising only the least desirable staple, e.g. millet in areas where the preference is for rice, or the lowest grade of staple; limiting the size of each transaction to a small quantity; and limiting the number of outlets to create long waiting times. Some care has to be taken with the relative values of food and labour as a situation may arise (as in Burkina Faso in 1984), where it was worthwhile for a better-off person to pay a poor person to stand in line for food.
There are few recent examples of market interventions to prevent food crises. An HEA assessment in northern Tanzania 1999, concluded that keeping down food prices would substantially reduce an assessed food gap (of approximately 30,000MT), i.e. cheaper prices would allow people to acquire sufficient food without substantial loss of household assets. In the event, this was achieved by the release of grain from the national food reserve to commercial traders. No direct relief distribution was done.
In some situations, it may be possible to select a commodity for food distribution which is more likely to reach poor people. While this may appear to be a useful method in situations where rich and poor people have different staple diets, its potential impact may be limited because richer people may be able to sell the commodity to obtain the food they really want to eat. It is now widely recognised that it is not appropriate to give people food which is unfamiliar and unacceptable (The Sphere project, 2004). Furthermore, it is often not possible to select commodities, as they are usually determined by global surpluses and food aid pipelines.
Food for work
Food for work programme in Ethiopia
Food for work (FFW), i.e. paying people in food for work, may be used as a means of targeting food to people who are in need, and/or to achieve a development objective which will benefit the community in the long or short term, e.g. building a road. FFW is extremely difficult to organise on any scale under crisis conditions, so it is rarely suitable as the primary food distribution mechanism in an emergency (Jayne et al, 2000 and Devereau, 2000). Appropriate work must be found, technical supervision and an administration arranged. There is also a need for non-food inputs (e.g. roadwork requires specialist tools, cement and other material for culverts, etc)5. The administrative demands involved are often unmanageable in an emergency context.
In the absence of a viable work objective, FFW effectively becomes a strategy for targeting food - the assumption being that people who do not require food will not take up employment. In this instance, the justification for the use of FFW becomes a moral one: people will become dependent on free food aid (and thereby fail to make the most of the economic opportunities available to them) if they do not work to receive it.
FFW is best suited to areas where there are adequate employment opportunities for the non-targeted groups (Jackson and Wickrema, 1998), where rates of pay are of a value equivalent to normal labour rates, and where a free food component is organised for households which are not able to participate in FFW.
Box 11 shows the important considerations when setting the pay rate and illustrates the importance of adequate local knowledge of household economies to inform the decision. It may not be possible to fulfil all the requirements in an extremely poor context and a combination of administrative and selftargeting is likely to be more appropriate (Sharp, 1997).
Box 11: Setting a wage rate for FFW
|Requirements of the wage
. Lower than the market rate
. Sufficient to meet
. Higher than the income
obtained through marginal /
damaging coping strategies
(such as firewood collection)
|Consequence if requirements
. Will not self select the poorest
. Basic needs will not be met
and destitution could
. Poor will not be able to participate
and only wealthier groups with
spare labour will benefit
Conclusions for best practice
- The best targeting systems employ several elements, rather
than relying on a single method. For example, in Ethiopia, the
Employment Generation Scheme is geographically targeted,
adopts a self targeting element for eligibility, but also
includes a free food component for the 20% of the population
that cannot participate in food for work.
- In determining eligibility, a judgement needs to be made
about which type of error is more acceptable; an inclusion
error or an exclusion error. It may, for example, be
necessary to allow for some diversion of food to authorities,
in order to ensure that the target group is reached, just as a
dry supplementary food ration is increased to take into
account household sharing. In other situations, it may be
necessary to incur exclusion errors in order reduce inclusion
errors. For example, in 1992 in Zimbabwe, Save the Children
was asked by the community to deliver food using a wet
supplementary feeding programme. This was regarded as
being the most transparent way to deliver food and the
community felt this method avoided corruption and diversion
- A priority in determining eligibility is to reach an acceptable
agreement between the community and the external agency
on the criteria. In the absence of this agreement, inclusion
and exclusion errors are likely to be extensive.
- It is not only the criteria, and their application, which is
central to targeting success. Commodity choice, distribution
method and the timing of the distribution can also contribute
to the achievement of targeting objectives.
Jaspars and Shoham,1999
3HelpAge International advises that "whilst chronological age provides a convenient means of defining a population group, it is often very limited as it does not reflect the understanding of ageing within a specific social or cultural situation. It is necessary, therefore, to determine culturally specific definitions of age and use them in conjunction with the UN definition of age (>60 years), to define context specific age" (HelpAge International, 2001)
4For example, the SADC Vulnerability Assessment Committee recommended two indicators: chronic illness of head of household, and households headed by the elderly, in particular those headed by women (SADC, 2003).
5Large scale projects are often designed with a five year preparation period (Jayne et al, 2000)
Taken from Field Exchange Issue 101, July 2004