Linking wealth and household head traits via quantile multilevel models in South Africa

Orientation: This study explored how household wealth in South Africa relates to key socioeconomic traits of household heads, against the backdrop of persistent inequality and growing scholarly interest in wealth as a driver of well-being and mobility.Research purpose: The study aimed to answer two central questions: (1) Are household head characteristics associated with different points of household wealth distribution across South African districts? (2) Is there greater variation in wealth within districts than between districts in these associations?Motivation of the study: Despite growing literature on wealth, few studies use micro-level data in developing countries. South Africa’s unequal context and the clustered nature of the National Income Dynamic Study (NIDS) data highlight the need for methods that capture distributional and geographic variation.Research approach/design and method: This study applied linear quantile multilevel modelling (LQMM) to Wave 5 NIDS data, accounting for district-level clustering and capturing how household head traits affect wealth across its distribution.Main findings: Household head characteristics – particularly age, education, marital status, gender and ethnicity – are significantly associated with household wealth. Importantly, these relationships vary across different quantiles of the wealth distribution, and there is substantial variation in wealth within and between districts.Practical/managerial implications: Given the heterogeneity in wealth outcomes, policies aimed at improving economic well-being in South Africa should consider both the geographic context (district-level disparities) and the distributional effects of household head characteristics. One-size-fits-all approaches may fail to address deeper inequalities.Contribution/value-add: This study advances the literature by using LQMM to model wealth across districts and distribution levels, emphasising district-level wealth disparities and deepening understanding of how socioeconomic traits shape wealth in unequal, post-apartheid South Africa. This model captures differences in effects across quantiles but does not correct for endogeneity from things such as omitted variables or measurement error.