Abstract:
This paper examines the household wealth construct as measured in NIDS Wave 4 (2014‐2015).
Questionnaire design changes in the wealth module between NIDS Waves 2 and 4 include the addition of questions on household possessions assets and the inclusion of a variable for private property versus communal property rights. For derived household net worth, the inclusion of household possessions assets reduces estimates of household inequality in Wave 4 compared to Wave 2, and alters the portfolio composition of household assets most markedly for low income households. A unique feature of NIDS Wave 4 is that it now allows for accurate identification of South Africa’s dual land tenure system. In Tribal Authority Areas (TAAs), households rarely have to finance the acquisition of a home partly because of communal property rights. This results in a very different portfolio of household liabilities than in other areas in the country, where real estate debt dominates the liability portfolio. When comparing NIDS aggregated national totals for estimates of household assets, liabilities and net worth with South African Reserve Bank (SARB) totals that use the national accounts, the NIDS Wave 4 data differs from the SARB data most significantly for financial assets, which are severely under‐estimated in NIDS. This is likely a result of disproportionately high attrition among high‐income households in the sample over time that cannot be compensated for sufficiently with survey weights. This suggests that it is urgent that the NIDS sample be refreshed, with an oversampling of high‐income households necessary.
