Comparing alternative measures of household income: evidence from the Khayelitsha/Mitchell's Plain survey

Working Paper


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title

University of Cape Town

Household income is a variable that is used widely for economic and sociological analysis. Little has been written about the optimal way to generate the information necessary to calculate household income. Most South African analyses use a household income variable generated by a single household respondent reporting on the household income. The Khayelitsha/Mitchell's Plain Survey provides a unique opportunity to explore alternative ways of generating this variable. We compare the estimates of household income obtained from the household module to estimates of household income obtained by aggregating the detailed income data from the adult module of the survey. We show that household income estimates for the KMP survey tend to be higher and to have greater variation when estimated by aggregating individual income data compared to the estimates obtained in the household module. The difference between income estimates has a material impact on the secondary analysis of data. This is illustrated through the use of Gini coefficients, a simple measure of income-inequality. Household income measured at the household level appears to underestimate household income-inequality in a sample.