Targeting Vulnerable Households Within the Context of HIV/AIDS in Malawi

Summary of evaluation¹

By Maja Munk and Dr. Neil Fisher

Maja Munk has been working in the food security sector for ACF/AAH in Africa and Asia since 2002. Her photographs have been published in periodicals and books in the Americas, Europe and Asia. She is currently working in Liberia as a programme coordinator.

Since 2002, Neil Fisher has been working in Malawi for AAH and is currently responsible for the Integrated Nutrition and Food Security Surveillance Programme. Previous work includes 12 years with universities in Kenya and Nigeria teaching Crop Production and Farming Systems and researching indigenous farming systems.

The HIV/AIDS epidemic in Malawi is receiving increasingly committed responses from its government, donors, religious institutions, the UN, and NGOs. Programmes addressing prevention, testing, treatment and vulnerability are gaining momentum. Currently many of these programmes aim specifically to target HIV/AIDS infected/affected individuals and/or households. While this may be an appropriate approach for many types of interventions (such as HIV-specific prevention, curative, medical, educational or nutritional programmes), for some food security inputs programmes this may not be the most effective approach. The majority of rural Malawians are chronically impoverished and malnourished; they exist on the edge, where a shock to, or within, an average rural household can push it to a degree of vulnerability that they may not be able to withstand.

Within this context, it may be more important to improve the selection of the most vulnerable households, whether the vulnerability is rooted in HIV/AIDS or not - first considering the degree and aspect of a household's vulnerability, and then considering the HIV status of the beneficiaries when developing the most appropriate intervention. By accurately targeting the more vulnerable households in countries, like Malawi, with high rates of HIV infection, infected/affected households will necessarily be targeted.

Action Against Hunger (AAH) examined the weaknesses and strengths of the current proxies used to identify HIV/AIDS infected/affected and/or vulnerable households in rural Malawi. After identifying the drawbacks of the proxies, AAH developed a methodology to mitigate their weaknesses. The methodology was piloted during an ECHO (European Commission Humanitarian Office) funded farm inputs distribution in Malawi's central district of Ntchisi.

A woman receives groundnuts as part of the programme

In Malawi, like many other countries, there is fear, ignorance, a lack of accessible free and anonymous testing, and a common assumption that all people tested for HIV have AIDS. Many women need their husbands' approval to be tested, and there is difficulty preserving privacy within the village context. Given the personal and social ramifications of knowing your status and/or being associated with the disease, few rural Malawians undergo voluntary counselling and testing (VCT) or are willing to be openly identified as being HIV positive. Until the impact of the current investments in VCT and antiretro viral (ARV) treatment are felt by Malawi's rural population, and the rural population is successfully sensitised to HIV/AIDS, identifying HIV infected/affected household will continue to be problematic. Additionally, there may be negative ramifications for households associated with programmes specifically targeting HIV/AIDS infected/affected families.

Correctly targeting vulnerable households is difficult in most contexts; accurately targeting households who are vulnerable directly or indirectly because of HIV/AIDS is, on any significant scale, not practically possible in rural Malawi. The substantial increase of VCT and ARV throughout the country may, in the future, make correct targeting of these households possible. However, currently donors and NGOs are increasingly attempting to address the HIV/AIDS pandemic by targeting HIV/AIDS infected/ affected households. In the absence of knowledge of an individual's HIV status, proxies are often used. The popular proxies, while theoretically indicative, are often in practice imprecise, and may not, in Malawi's context, be the most effective tools to select the HIV/AIDS infected/affected vulnerable population.

Proxies

In Malawi, there are commonly used proxies (such as mortality/morbidity, orphan, female/ child /elderly-headed households, and high depend ency ratio households) that are associated withHIV /AIDS and vulnerability. For a variety of reasons, these proxies may be problematic.

Common proxies

'Chronic illness' is a popular proxy often associated with HIV, but the term is often not well defined. Sometimes 'chronic' is given a time scale (i.e. three months, one year, etc.), but commonly the definition of 'illness' is left assumed, rather than defined. Conversations with villagers indicate that people have a very varied idea of what 'chronic illness' means - even within a given timeframe. At the village level, a chronically ill person can be someone suffering from a bad back or arthritis, or someone who has had several bouts of different related or non-related ailments within the specified time. While a welldefined 'chronic illness' may capture household members who have been unhealthy for a specific period of time, it can be an imprecise proxy for capturing HIV/AIDS or its opportunistic illnesses. There are also many types of chronic illnesses historically existing in southern Africa that may not have any relation to HIV/AIDS or its associated diseases. However, 'chronic illness' may be a good indicator of household stress due to a possible decreased labour capacity and/or increased expenditure.

The recent death of a household member is also a common proxy. Unlike 'chronic illness', 'death' is well defined. Historic data support the view that the presence of HIV/AIDS has lowered Malawi's life expectancy. However, one must take into account the person's age and reason for death before the death is associated with HIV/AIDS. While the presence of death does not necessarily indicate the presence of HIV/AIDS, if the person who died was a productive member of the household, 'death' may indicate increased household stress.

Like chronic illness, the term 'orphan' lacks a precise definition. While NGOs commonly assume an orphan to be a child with one or both parents dead, at the village level the definition of 'orphan' is not so clear-cut. Numerous individual and group discussions revealed that central rural Malawians have a wide spectrum of opinions when it comes to defining an orphan. The varying definitions fall into two basic schools of thought: an orphan is a child who has one or two parents dead irrespective of the support the child receives from its remaining family; or an orphan is a child who does not have access to basic necessities, whether its parents are alive or not, and may rely on people other than its parents to satisfy those needs. Even within a household, there can be different options regarding the status of a child. A woman's second husband may consider the children from his wife's first marriage to be orphans, the wife may or may not consider her own children as orphans.

If an orphan is defined by the death of one or both parents, there is also the question: once an orphan always an orphan? That is, if a child is considered an orphan because one parent died, does it remain an orphan after the remaining parent remarries? Does an orphan continue being an orphan after they get married or on their 16th or 18th birthday? Or is being an orphan restricted to a state of being dependent on others for your needs? If a grandmother is taking care of her grandchild, how much support is the child's parent/s responsible for, before the child is considered an orphan? While the answers to these questions may, on an abstract level, seem straightforward, at the field level they are often complicated.

The presence of 'orphans' in a household is not necessarily a reliable proxy for HIV/AIDS in Malawi. While governmental and non-governmental actors have developed definitions for 'orphans', often these definitions do not translate into field realities. It should not be automatically assumed that the presence of an 'orphan' necessitates a death in the child's family (as one, both or neither of the child's parents may be dead), or that the presence of the 'orphan' necessarily indicates a long-term economic burden on the hosting family (as the child may be receiving other support, the presence of an older 'orphan' may represent a positive or neutral contribution to a household, or the child may be hosted for a short period of time). However, the hosted orphan may be at a higher risk of vulnerability due to inequality of care, than the other children in the household. That said, specifically targeting the orphan, as oppose to the household as a whole, may further marginalise the child.

The proxy 'female-headed household', as opposed to male-headed, assumes the absence of an active adult male contributing to the house. The absence of a male-head has a compounded effect: it puts more of a burden on the female-head as the primary provider for the household, often in an environment where women have fewer income generating opportunities than men. AAH surveys in central Malawi have found that polygamy in rural central Malawi is quite widespread. However, the scope of polygamy does not readily present itself, as the polygamy practised in central Malawi is matrilocational, i.e. the male-head divides his time between the households of his co-wives, as opposed to sharing a common compound with them. In this structure, each co-wife's house lacks the full contribution of the husband or the support of the other co-wives. Interviews with co-wives found that these households can effectively be female-headed, as the wife has to provide for the household with, most commonly, little or no support from the husband. However, these households can present themselves as male-headed as there is a male head of house, even if only figuratively.

Additionally, in a culture where the man controls household expenditure, a male's presence can be negative if the priorities of the man contradict the priorities of the household's food security - a common complaint of village women. Therefore, it is possible for a household to be male-headed but the male's contribution to the household to be neutral or negative. What may matter more than the sex of the head of house is the number of active contributing adults, and to what degree they positively contribute to the household.

'Elderly' and 'child-headed' households are likely to be vulnerable as they are often without a fully active contributing adult. The degree of their vulnerability lies both in the household head's ability to contribute to the house and the amount of support that they receive from their social-safety network. This can be a particularly important resource for elderly- headed households. Often the elderly live in close proximity to their adult children, however, the support provided can vary widely. Child-headed households may have to rely on a smaller family support structure (as they have no adult children to rely on). It is also important to remember that the threshold for childhood varies culturally in Malawi, it is not uncommon for teenagers to marry and establish their own households before their 18th birthday. While these households may be technically child-headed, to call them such may be misleading. While elderlyheaded households with children are frequently associated with HIV/AIDS, the presence of a generation gap in a household is not unusual where middle age groups migrate for work. This predates the onset of HIV/AIDS and should not automatically be associated with HIV/AIDS. However, as the family support structure in Malawi shifts focus from the extended family network towards the nuclear family, child and elderly-headed households may become increasingly disenfranchised.

Developing alternative proxies

Although measures are underway to step-up VCT, widespread HIV testing of rural Malawians will remain problematic in terms of ethics and plausibility. Therefore, developing empirical proxies is necessary. The reason why HIV/AIDS infected/affected households are targeted is because the presence of the disease is believed to increase vulnerability. Therefore, one way of indirectly identifying HIV/AIDS infected/affected households is to identify a household's degree and aspect of vulnerability.

Calculating a household's dependency ratio is a common way to determine a household's degree of vulnerability. In the classic dependency ratio, the number of dependent members is expressed as a ratio to the productive members (Dependency Ratio=100*number of dependents/number of productive adults). The problem with this equation is that a person is valued as all (productive) or nothing (dependent). This rating of contribution does not reflect the gradient of contribution over a lifetime.

AAH explored an alternative measure of dependency that allows for varying degrees of contribution, and developed a 'food/labour dependency ratio'. Knowing the age and gender of each household member, one can calculate the theoretical daily food energy need and potential labour capacity of a household.

When calculating the potential contribution that an individual makes towards generating cash and/or kind, an adult (21-55 years) male was rated at 1.0 (as his workload is primarily limited to income earning and/or crop production) and an adult female at 0.8 (as her workload consists of not only income earning and/or crop production, but also household work, child rearing, caring for the ill, etc.). The positive contribution of children starts at 12 years old (rated at 0.1 for males and 0.08 for females) and increases until 20 years, when they become a full adult. Then their contribution decreases from 56 years of age until 68 years of age (when they are valued at 0.09 for males and 0.02 for females), after 68 years they are rated at zero.

Similarly, the food energy requirements of each household member is calculated on a sliding scale by sex and age, starting at 1300kcals for female and male one-year olds and ascending to 2900kcal for lactating active adult females and 3300kcals for active adult males. By dividing a household's food energy requirement by its labour potential, the 'food/labour ratio' is formed. At 5,800 kilocalories per productive member, an average family of two active and three inactive members in Malawi is generally food secure. As the active household members become responsible for more than 5,800 kilocalories, the household becomes increasingly stressed.

Targeting in Ntchisi district

Beneficiaries were selected on the basis of household food/labour ratio and degree of material poverty. Chronically ill adults² were assumed not to be contributing to the labour of a household. Dependency was then scored from 1 (lowest) to 10 (highest). The poverty score was created by placing a monetary value on productive and non-productive assets and calculating their total value. The total was then divided by the number of people within the household to formulate a per-capita value, and ranked from 1 (poorest) to 10 (richest). The dependency/poverty score was then created:

10+Dependency Score-Assets Score =
Dependency/Poverty Score

The score took values from 1 (least vulnerable) to 19 (most vulnerable). After this score was ascertained, one point was added for each orphan³ in the house. Two points were added to households that had elderly people caring for children. Households that scored 11 or more points were chosen as beneficiaries.

A short, ten-minute, targeting questionnaire capturing the essential data needed (household structure, mortality/morbidity, assets, and land holdings) for the targeting methodology was conducted in four health centres and their surrounding villages, within the catchment. Of the 3,500 people interviewed with the questionnaire, 64% of the total number of households qualified under the criteria used in the targeting methodology. Due to time constraints, the remaining 36% of beneficiaries were selected through local leaders using defined criteria: female-headed households, elderly-headed households with children, and households with orphans. Household characteristics determined the inputs received: vulnerable households with sufficient land and labour received crop inputs, vulnerable households with insufficient land or labour received chickens and feed.

Evaluation of the methodology

Constraints

The effect of the piloted targeting methodology was diluted, and AAH's ability to evaluate it was somewhat limited, as, due to time constraints, not all beneficiary households were targeted through the food/labour dependency ratio. Additionally, AAH liased with local leaders to inform selected households and ensure that they collected their beneficiary cards at designated points before the distribution. Follow-ups found that some of the cards were given to people other than the intended beneficiaries.

During the monitoring and evaluation process, it was not possible to determine under which methodology a particular household was targeted. It was, therefore, not possible to ascertain during the evaluation process if incorrectly targeted households were selected by the piloted methodology or selected by local leaders. As an unknown third of the beneficiaries were not selected through the piloted methodology, properly evaluating its ability to correctly identify vulnerable households was difficult.

Selected households

Of the surveyed beneficiary households, 65% qualified on one criteria count and 17% qualified on two criteria counts. 18% should not have qualified as they were in the richest two quartiles and in the two quartiles with least dependency.

Despite the fact that the potential effectiveness of the food/labour dependency ratio may have been diluted as 36% of the beneficiaries were selected via local leaders, the methodology was able to skew the targeted households towards the higher food/labour dependency households when compared to the base population. The only exception was the proportion of beneficiaries selected in the >8000kcal group, which is slightly less than the base population. This underrepresentation may be caused by these disenfranchised households not accessing health structures and not having a voice in the community, and therefore less likely to be picked-up during the selection process.

In the food/labour ratio (figure 1), the threshold for vulnerability is calculated at 5800kcal per man equivalent. The targeting methodology was able to select proportionally more households that have a high food/labour dependency. In the dependency ratio, household vulnerability increases as the ratio increases from 100 -the methodology was able to proportionally select more of these high dependency households as well (see figure 2).

Households which were not selected by the questionnaire had a food/labour ratio of 4,825kcal per man equivalent. The 17% of targeted households who should not have qualified had a food/labour ratio of 5,056kcal per man equivalent. Qualifying households had a food/labour ratio of between 6,321 and 8,000Kcal per man equivalent.

There were statistically significant demographic differences between targeted households and the base population. Compared to the base population, the piloted targeting methodology picked up proportionately more households headed by persons aged 31 years or older, headed by women, and/or headed by widow/ers. These households had fewer productive and non-productive assets than the base population. The presence of orphans in targeted households was between 2% and 4% higher than the base population. The poorest and most dependent 17% of the sample was more likely to be female-headed households with orphans; if these households had orphans they were more likely to have a higher number of orphans per household than the other segments of the sample.

Estimates of inclusion/exclusion errors

The piloted methodology had an estimated inclusion error of between 21% and 38% and exclusion error of <49%. The inclusion errors appear to be better than those of the government of Malawi's DFID⁴ funded Targeted Input Programme (TIP) in 2000-01. There are a number of possible reasons for this large inclusion error: the 36% who were selected through local leaders, rather than through the targeting interview, may not have conformed to the dependency/poverty criteria that were used for the 64% selected by the piloted methodology; the correct people were interviewed for the beneficiary profile survey, but they gave different answers to the household structure and asset questions during the survey than during the targeting interview⁵; some beneficiary cards were misallocated. If these assumptions are true, then the true inclusion error lies somewhere between 21% and 38%, as the portion attributed to households giving differing answers is, in part, a measurement error rather than a real inclusion error. While the inclusion error of the piloted methodology is the same or worse than the inclusion error for AAHs distribution in Ntchisi's neighbouring district of Kasungu (see table 1), where a traditional method of targeting through local leaders was used, the exclusion error is better. This is important as exclusion errors are arguably more important than inclusion errors, as they can trigger resentment within the concerned communities.

The quantitative inclusion and exclusion errors, as well as the qualitative input of the enumerators and AAH food security staff who carried out the evaluation, indicate that within Malawi's context of widespread endemic poverty, it is difficult to differentiate the subtle degrees of poverty or dependency among poor households to find the most vulnerable amongst the vulnerable. However, as most interventions do not have the capacity or mandate to target all of Malawi's impoverished households, selection criteria are needed. While the methodology piloted in Ntchisi is a step in the right direction, more subtle targeting tools will need to be developed. While identifying the 20 to 30% "least poor" is reasonably simple, differentiating between levels of poverty in the remaining 70 to 80% is much more difficult.

Conclusions

It is not HIV/AIDS itself that threatens a household's livelihood, but the ramifications of the disease that make the household increasingly vulnerable as their income and expenditure are skewed, and as the ratio of active members to dependents shifts. Perhaps within Malawi's context of widespread chronic poverty, it is less essential for NGOs implementing non-HIV specific food security interventions to find households infected/affected by HIV/AIDS than to identify households that are vulnerable. The nature of the vulnerability should be taken into account when designing programmes. Once the most vulnerable households have been correctly identified, the specific stresses of HIV/AIDS can be considered. Due to the limitations of proxies being applied practically in rural Malawi, humanitarian actors should not use them as the simple solution to identify beneficiary households and should question if using them is, in fact, appropriate for their intervention. While the targeting methodology developed by AAH was a forward move, methodologies to improve identification of the most appropriate vulnerable households to target should be further developed and explored.

For further information, contact AAH Malawi, email: aah@globemw.net

¹Munk M and Fisher N (2004). Targeting vulnerable households within the context of HIV/AIDS. An evaluation of a piloted methodology. Ntchisi, Malawi. Action Against Hunger

²For the questionnaire, 'chronically ill' was defined as ill for more than three months

³ For the questionnaire, orphan was defined as under 18 years old with both parents dead

⁴UK Department for International Development

⁵Targeting interviews were not conducted in the home, impact assessment interviews were

⁶Nyirongo CC, Msiska FBM, Mdyetseni HAJ, Kamanga FMCE, Levy, Sarah (2003). 2002-03 Extended Targeted Input Programme: Evaluation module 1: Food production and secu rity in rural Malawi. Malawi Ministry of Agriculture, Irrigation and Food Security and UK Department for International Development.