Volume 37 - Article 39 | Pages 1297–1326 Author has provided data and code for replicating results

Decomposing changes in household measures: Household size and services in South Africa, 1994–2012

By Martin Wittenberg, Mark Collinson, Tom Harris

Print this page  Facebook  Twitter


Date received:28 Jun 2016
Date published:20 Oct 2017
Word count:6287
Keywords:decomposition, electricity access, household formation, household size, panel data, South Africa
Additional files:readme.37-39 (text file, 2 kB)
 37-39_Online-Appendix (Word file, 14 kB)
 demographic-research.37-39 (zip file, 7 MB)


Background: Household trends are generally tracked by means of repeated cross-sections, such as censuses or nationally representative surveys. However, the trends may be driven either by changes within households over time or the way in which the processes of household formation/dissolution interact with the measure in question.

Objective: We aim to develop a method that enables us to apportion changes in a household measure to changes that happen within households and changes that occur due to household formation and dissolution. In particular we intend to show how South African households have reduced in size and how access to services has increased.

Methods: We develop a formula for decomposing a household outcome measure. We apply the formula to household size and electricity access data from the Agincourt health and demographic surveillance site for the period 1994 to 2012. We also apply it to the National Income Dynamics Survey of South Africa from 2008 to 2012. We compare the results to the pattern derived from nationally representative surveys run by Statistics South Africa since 1994.

Results: The overall reduction in household size is fuelled by rapid household formation, much of which is intertwined with shifts in location. Access to services has been reduced by the process of new household formation. Neither finding is evident from cross-sectional data.

Contribution: We introduce a new decomposition technique which can be used with longitudinal data and discuss the insights that it provides.

Author's Affiliation

Martin Wittenberg - University of Cape Town (UCT), South Africa [Email]
Mark Collinson - University of the Witwatersrand, South Africa [Email]
Tom Harris - University of Cape Town (UCT), South Africa [Email]

Other articles by the same author/authors in Demographic Research

» Household structure vs. composition: Understanding gendered effects on educational progress in rural South Africa
Volume 37 - Article 59

» Human capital on the move: Education as a determinant of internal migration in selected INDEPTH surveillance populations in Africa
Volume 34 - Article 30

» The age pattern of increases in mortality affected by HIV: Bayesian fit of the Heligman-Pollard Model to data from the Agincourt HDSS field site in rural northeast South Africa
Volume 29 - Article 39

» The implications of long term community involvement for the production and circulation of population knowledge
Volume 17 - Article 13

Most recent similar articles in Demographic Research

» Ultra-low fertility in South Korea: The role of the tempo effect
Volume 38 - Article 22    | Keywords: decomposition

» Household structure vs. composition: Understanding gendered effects on educational progress in rural South Africa
Volume 37 - Article 59    | Keywords: South Africa

» The wage penalty for motherhood: Evidence on discrimination from panel data and a survey experiment for Switzerland
Volume 37 - Article 56    | Keywords: panel data

» The compositional effects of education and employment on Greek male and female fertility rates during 2000‒2014
Volume 36 - Article 47    | Keywords: decomposition

» The long-term determinants of marital fertility in the developed world (19th and 20th centuries): The role of welfare policies
Volume 36 - Article 42    | Keywords: panel data