Measuring household food insecurity in emergencies: WFP's Household Food Consumption Approach

Measuring food security continues to challenge the humanitarian community. While internationally recognised indicators and standardised anthropometric measurements exist to assess the prevalence and severity of malnutrition, equivalent indicators and procedures are not available to assess the extent and severity of household food security. Instead, a variety of indicators and approaches is used to describe the multi-faceted dimensions of food insecurity and the status of household food availability, access and utilisation. This diversity of indicators and approaches makes it difficult to compare the food security situation across settings, population groups and time, and to prioritise the allocation of limited resources. Over the past few years, WFP has embarked on several initiatives to improve methods to assess household food security in emergencies. This article explores one such initiative, which focuses on dietary diversity, food frequency and food sources as a proxy indicator for food security. This work was undertaken within the framework of the WFP Strengthening Emergency Needs Assessment Capacity (SENAC) project, with funding from DFID, ECHO, GTZ, CIDA and the Danish government.

The need for standardisation of food security measurement

For years, malnutrition indicators based on the prevalence of stunting, underweight or wasting among children under five years of age have been employed as the programming foundation and criterion for food aid interventions. But the prevalence of malnutrition does not necessarily reflect the level of household food security, and cannot be considered an adequate proxy indicator. This is because malnutrition is not always caused by household food insecurity. As described in UNICEF’s model, there are three major underlying causes of malnutrition, of which household food insecurity is just one (the other two are insufficient health services and an unhealthy environment, and inadequate care practices). Furthermore, malnutrition is a late indicator of food insecurity because the child is malnourished by the time the problem is identified. A household food security indicator would allow early identification of the problem prior to such a deterioration in nutritional status.

One of the main problems with measuring household food security is the absence of a single indicator that could capture the definition of ‘food-insecure households’. The definition of food security adopted at the World Food Summit in 1996 is comprehensive, but rather complex and ambiguous:

Food security, at the individual, household, national, regional and global levels [is achieved] when all people, at all times, have physical and economic access to sufficient, safe and nutritious food to meet their dietary needs and food preferences for an active and healthy life.

This definition is not very helpful in measuring the proportion of food-insecure households because it does not include clear thresholds, and because it conflates different levels (individual, household, country and international). Hence, the results of food security measurement may vary according to who conducts each assessment.

The humanitarian community has developed different methodological approaches to measuring household food security. These range from elaborate methods such as the household economy approach to a simple judgment-based approach. While efforts to systematise these methods have improved the quality and transparency of analytical outcomes, they are still insufficient because different methods result in different estimates. As a result, assessments still lack comparability across countries or over time. Without a means to ensure comparability, the humanitarian community will continue to have difficulty making effective, efficient and accountable decisions on funding and priorities for food and other aid. This lack of comparability will also impair communications with, and advocacy to, donors and the general public on behalf of countries or populations most in need of assistance.

Household food consumption as an indicator of food security

To contribute to efforts to standardise household food security measurement, WFP has explored the use of an indicator that could adequately estimate the severity of household food insecurity, and thus indicate the potential urgency of intervention. This new indicator could also be used as the basis for determining the number of households to be surveyed, in order to have statistically representative data on food security.

A score of dietary diversity and food frequency, derived from information about households’ consumption of specific food items during a designated period, could form the basis for one such indicator. Its main advantage is objectivity and measurability. Several recall periods can be envisaged, and the number of food items asked about can vary, depending on the local dietary culture and the purpose of the assessment. The International Food Policy Research Institute (IFPRI) and the Food and Nutrition Assistance Project (FANTA) have conducted a number of studies that examined the relative merits of various approaches. In its recent emergency assessments, WFP has typically applied a seven-day recall time, and focused on a selected number of major food groups. The scores reached are combined with other variables to determine household food security groups.

WFP’s pilots in the 2004 and 2005 Darfur Emergency Food Security and Nutrition Assessments

While an approach based on dietary diversity and food frequency was used in several baseline food security and vulnerability analyses conducted by WFP, its application in an emergency assessment was pioneered in Darfur, Sudan, in 2004. In this assessment, the approach combined three elements: (i) dietary diversity, defined as the number of unique food items (such as pulses, sorghum, meat and sugar) consumed during the previous seven days; (ii) food consumption frequency, defined as the number of days for which each food item was consumed over the previous seven days; and (iii) the primary source of each food item.

The proportion of food-insecure households was estimated in two steps:

1. Households were classified into three food consumption groups (‘acceptable’, ‘borderline’ and ‘poor’) according to the diversity of the diet and consumption frequency.
2. Depending on the primary source of each food item, specifically whether it was from food aid, households were further classified into three food security groups (‘food-secure’, ‘vulnerable to becoming food-insecure’ and ‘food-insecure’). This step was aimed at estimating the sustainability of the current food consumption level through an analysis of the primary source of the foods consumed.

The methods used for the 2004 assessment were further refined in the next assessment in Darfur, in 2005. The methods used in 2004 assumed that households with acceptable food consumption, but which were dependent on food aid, were vulnerable to becoming food-insecure. However, high- and middle-income households placed in this group might have saved substantial resources by receiving food aid (i.e. a possible ‘inclusion error’), or they may have had the capacity to spend money on food or to draw from their own food stocks.

To address the possible misclassification of these households, weekly per-capita expenditure on food was added to the cross-tabulation steps in the 2005 assessment. By introducing the threshold for quintiles of weekly per capita expenditure on food, it becomes possible to exclude households that enjoy acceptable food consumption through food aid. Households with borderline or acceptable food consumption patterns were considered food-secure, vulnerable or food-insecure according to their degree of reliance on food aid, and the level of their weekly per capita expenditure on food.

Challenges for moving forward with a Household Food Consumption Approach 

Despite the methodological innovations made in the two assessments in Darfur, major issues still need to be tackled to make the best use possible of the Household Food Consumption Approach.

1. Household classification. One of the greatest methodological challenges is to establish common and absolute thresholds for classifying households into food consumption groups (such as ‘acceptable’, ‘borderline’ and ‘poor’) across all countries and situations. Principal component analysis and cluster analysis used in Darfur resulted in different thresholds derived from the dataset of the respective assessments. Therefore, there are limitations to the comparisons that can be made between the proportion of food-insecure households in the 2005 and 2004 assessments, and between other assessments in different countries.
2. Seasonality. The reliability and reproducibility of households’ responses on food consumption frequency during the previous seven days can be questionable. However, this period of time may not adequately account for variations according to the period when the survey is taking place (i.e. seasonality). Food consumption and food sources are likely to vary depending on the proximity of the harvest, periods when labour opportunities provide access to cash income and the timing of food aid distributions.
3. Sustainability of consumption. The approach followed in the Darfur assessments uses the consumption of food aid as a major criterion to classify household food security groups. In countries or areas where food aid distributions are not implemented, other variables would be required to address the sustainability of the current food consumption pattern, for example by taking into account other sources of food (such as loans or gifts) as indicators of an unsustainable pattern, as well as complementary information on sources of income, the use of assets and coping strategies. However, these variables may be context-specific, and again this limits the comparability of results across settings and over time.
4. Intra-household distribution. The approach as applied in Darfur does not provide information on variations in food consumption within the household. There may be cases where the food consumption pattern is acceptable at household level, but poor for some members of the household. A household may qualify as food-secure on the basis of its food consumption pattern, and may be judged to be in no need of assistance, while some support may in fact be required in the form of targeted food aid (e.g. supplementary feeding, school feeding) or non-food interventions (e.g. behaviour change communication on care practices). Additional information would thus be required, in particular on the typically vulnerable, such as children under five years of age, pregnant and lactating women, the elderly and the chronically sick, as well as on any other vulnerable individuals according to the context, including their individual food consumption pattern and health and nutritional status (anthropometric measurements).

Rather than being viewed negatively, these and other limitations should be addressed by using these methods in a large variety of emergencies, testing different thresholds and repeating the surveys over time among the same population groups, to assess seasonal variations. In addition, further analyses of the correlations between dietary diversity, food consumption frequency and food sources and other key food security-related indicators, such as livelihood activities, nutritional status and food availability outlook, should be conducted to estimate the capacity of this approach to assess the various dimensions of household food security.

WFP welcomes collaboration with other humanitarian agencies involved in food security and nutritional assessments. This collaboration would help harmonise approaches and contribute to standardising food security measurement more broadly. The vast number of people suffering from food insecurity remind us of the urgency to improve the measurement of food insecurity and the way it is communicated to donors and to the public for advocacy, fundraising and intervention purposes.

Hirotsugu Aiga, MPH, PhD, is a senior programme adviser at Emergency Needs Assessment Services, WFP, Rome. His email address is: hirotsugu.aiga@wfp.org. Agnès Dhur, PhD, is an assessment methodology specialist in the Strengthening Emergency Needs Assessment Capacity (SENAC) project at WFP. Her email address is: agnes.dhur@wfp.org.