Poverty in African Households: the Limits of Survey and Census Representations
Measuring poverty in Africa and understanding its determinants and dynamics have become increasingly important with the targets set by the Millennium Development Goals (MDG) and the post-MDG agenda. There are diverse approaches to studying poverty: it can be measured as relative or absolute (Kakwani & Silber, 2008) and conceptualised as multidimensional or fuzzy (Alkire & Roche, 2012; Qizilbash & Clark, 2005). Although analysis may be undertaken at individual, household, district, national or regional level, understanding determinants and outcomes of poverty has largely focused on the household level. In countries with minimal state support of the poor and vulnerable, the domestic (that is, co-
residential and usually familial) group is the primary source of support, socialisation and resources. Although the really poor often live outside households (homeless, street children), few data capture and measure their well-being, because not being attached to households makes them statistically invisible. Many researchers see poverty, particularly persistent poverty, as fundamentally a householdlevel problem (Barrett, Carter, & Little, 2006). Livelihood approaches construct the household as ‘a site in which particularly intense social and economic interdependencies occur between a group of individuals’ (Ellis, 2000, p. 18). Thus, data for poverty analyses need to be collected on households. This article focuses on difficulties and dilemmas in doing this where survey and census definitions of ‘household’ encompass a notion of a bounded, largely impermeable, unit. We examine the limits of the survey approach when such a unit is applied, the implications for variables such as household size, and the ramifications for data analysis and thus understanding poverty and its determinants. We consider
what sorts of poverty-related issues may be missed or inaccurately represented because of the ways ‘household’ is defined in surveys. Anthropological critiques of economic well-being using the notion of ‘household’ are well-
established. Guyer and Peters demonstrated that an economic concept of household mapped poorly onto African social and economic organisation (Guyer, 1981; Guyer & Peters, 1987). They emphasised three key conceptual principles: (1) African households are not discretely bounded groups and different household members can draw on different personal networks to access resources; (2) households are not fixed forms but constantly evolving; and (3) households are differentiated along lines of gender and generation. These principles continue to apply today, and Guyer and Peters’s perceptive analysis is an essential backdrop to this article. Despite these longstanding anthropological insights into the nature of African households, much quantitative data collection and many analyses choose not to take account of them.