This paper investigates the validity of household survey data published by Statistics South Africa since 1993 and later integrated into the Post-Apartheid Labour Market Series (PALMS). A series of statistical adjustments are proposed, compared, and applied to primary data with the purpose of generating time-comparable, unbiased estimates, and accurate standard errors of labour earnings inequality coefficients. In particular, corrections deal with outliers and implausible data records, missing observations, bracket responses, breaks in the series, under-reporting of high incomes, and quarterly frequency. This work lays the ground for future research on the redistributive dynamics of economic policy in South Africa, which notably suffers from the presence of spurious shifts in repeated cross-sections.
income inequality distribution heterogeneity survey data imputation