Differences in food insecurity between adults and children in Zimbabwe

Summary of published research¹

Mother and child harvesting sweet potato on communal land in Zimbabwe

A variety of methods have been utilised to assess food insecurity, including measures based on national food supplies and anthropometric methods. Generally, assessments of food security require categorising the entire household's food security status based on responses from the household head. In the process, the views of other household members - most notably those of children - are not considered. This is an important omission as there is evidence of age bias in resource allocations in developing countries.

Neglecting the assessments of food intake of other household members may skew estimates of overall food security, food security within different groups, and analyses of the determinants of food security. A recent paper reports on a large scale survey from Zimbabwe where information regarding the food insecurity situation of the household was obtained from both a parent and a child. Three central questions were addressed in the survey:

• How does the overall incidence of food insecurity differ whether the child or the adult is the respondent?
• Within households, what is the correlation between adult and child reports of food insecurity?
• What are the determinants of adult and child reports of food insecurity?

The data used in this study were taken from a survey of over 6000 households across Zimbabwe in 2004. The survey was conducted by Catholic Relief Services (CRS) with funding from the US Agency for International Development (USAID). The sample frame was derived from the 2002 national census and comprised five districts, selected to represent the five basic areas of community life in Zimbabwe: urban, peri-urban, rural, commercial farm, and resettlement. In each district, a sample of households were selected by taking a sample of district wards, a sample of villages within each ward, a random sample of enumeration areas (EAs) within each village and a random sample of households in each selected EA. Finally, a household was retained in the sample if it contained a child between 6 and 18 years of age.

An adult in each household was asked demographic and economic questions pertaining to the household. In addition, a child between the ages of 6 and 18 years was randomly selected from those children present in the household to answer several questions. Both the child and the adult were asked the following questions regarding their food intake:

• How often do you have enough food? (Responses were always, sometimes, rarely, never)
• How many meals did you eat yesterday? (Responses were 0, 1, 2, 3, >3)

These two measures reflect individual intake of food within the household.

As both the adult and the child lived in the same household, their perceptions of food intake are likely to be correlated, even if they are not identical. As a result, a bivariate probit model, which allows for correlated errors between the equation for the adult and the equation for the child, was estimated in order to capture these correlations.

Results

Over half (57.1%) of the responses signified differences in reported food inadequacy for the adult and child in the same household. Among the cases where there were differences, 58.8% of the differences indicated a better situation for the child. Just under half (45.6%) of the adults and children reported a different number of meals consumed the previous day. As with the food inadequacy measure, most of these cases (59.1%) were in favour of the child (i.e. the child consumed more meals than the adult).

The information demonstrates that there are differences in adult and child reports of food insecurity using both measures. In particular, there are statistically significant differences in adult and child reports between perceptions of number of meals consumed and food inadequacy for adults and children in these households. More specifically, the correlation between adult and child reports of food inadequacy is 0.133, while the correlation between the adult and child reports of the number of meals measure of food insecurity is 0.107. With respect to child characteristics, the sample is evenly split between boys and girls. About 33% of the sample were orphans and the children were roughly evenly distributed across age groups. In terms of household characteristics, the vast majority of caregivers were female. Most adults worked in the farming sector, less than half of the homes were in good or fair condition, the average household size was approximately five people, and most individuals were interviewed at home.

Bivariate probit results for both measures showed that the correlation between adult and child reports was positive (q = 0.181 for food inadequacy and q = 0.232 for 'the not enough meals measure') and statistically significant. Thus, even after controlling for several observed factors, adult and child reports were still positively correlated. While the positive and significant correlation between adult and child reports of not having enough meals and food inadequacy may suggest that the food insecurity status of the adult is representative of child food insecurity status, the results from the bivariate probit analyses do reveal differences in food inadequacy and number of meals among members within the household.

Using the food inadequacy measure, there appears to be age preference in food distribution with a tendency to protect younger children against food insecurity. A child in the age group 6-9 years, when compared to a child aged 16-18 years, has a lower likelihood of reporting food inadequacy. In addition, all other child characteristics have no impact on adult or child reports of food inadequacy, with the exception of being a male orphan. Adults living with a male orphan are significantly less likely to report being food inadequate. Covariates that proxy for income and assets of the household (sector of work and housing quality, respectively) and represent greater economic resources are consistently negatively associated with both adult and child food insecurity.

Results from the 'not enough meals' measure are similar to those reported above, as both adults and children in homes with greater economic resources are less likely to report food insecurity. There are however a few differences. For instance, using this measure, children aged 6-9 years were more likely to report eating fewer meals, as were female orphans.

There are marked differences in the predicted probabilities of food insecurity for children and adults based on their demographic and socioeconomic characteristics. Children are significantly less likely than adults to report food insecurity across most household and child characteristics. However, most of these significant differences in food insecurity reports occur mainly for the food inadequacy measure of food insecurity. Starting with age, the results show a tendency to protect younger children, as younger children are generally predicted to be significantly less likely to report being food insecure than the adult in the same household across the age gradient. This finding is important mostly for the food inadequacy measure. The singular exception occurs for children aged between 13 and 15 years who are significantly less likely to report food insecurity by the 'not enough meals' measure. While there is evidence of children being protected along the age gradient, orphans do not appear to receive the same treatment.

Non-orphans, regardless of gender, are significantly less likely to report food insecurity across both measures. However, female orphans appear to be protected only by the food inadequacy measure and not by the 'number of meals' measure, while male orphans are not protected at all. Also, the gender of the caregiver is very important, as children in households with female caregivers are also protected against food insecurity. With regard to household size, children in most households are significantly less likely to report food insecurity by the food inadequacy measure, with the exception of children with eight household members who do not appear to be protected by either measure.

The variables that proxy for income (the sector of adult employment) and assets (quality of housing) generally follow the same pattern, in that children are less likely to be food insecure than adults. The notable exception is households where the adult works in the formal sector. In these households, for the food inadequacy measure, children are predicted to be more likely than adults to report being food insecure. In all other sectors of employment, there is evidence once again of protecting children. In fact, it appears that the poorer a family is based on the adult's employment sector, the greater the probability that the adult will report food insecurity when the child does not. These differences are significant at 1% for the food inadequacy measure. In particular, in the poorest homes, where the adult works in the labour/piecework sector, children are significantly less likely to report being food insecure than adults across both measures. For the other measure of household economic welfare - the quality of housing - the results show that in households residing in higher quality homes, children are significantly less likely than adults to report food insecurity across both measures. From these results we can conclude that children are more likely to be protected from food insecurity in households that reside in higher quality housing, but not in households where the head earns a relatively high income.

There are several implications of this paper for policymakers and practitioners in Zimbabwe and elsewhere. The presumption that all household members have the same level of food insecurity is false for different demographics within the household. While there are differences in reports of food insecurity, in the main, the children in the household have lower levels of food insecurity than the adults. Thus, efforts in Zimbabwe to reduce food insecurity among children that distribute benefits at the household level are likely to benefit children. This general finding does not hold for all groups, though, insofar as children in households with an adult working in the formal sector are more likely to be food insecure than adults. To reach children in this category, one strategy is to provide meals to vulnerable children in school settings. Such a policy would have the secondary benefit of encouraging school attendance, while also leading to improved child educational and health outcomes.

¹Kuku O, Gundersen G and Garasky S (2011). Differences in food insecurity between adults and children in Zimbabwe. Food Policy (2011) in press