Inequality in South Africa: what does a composite index of wellbeing reveal?

Policymakers need better information regarding wellbeing inequality to ascertain the contributing factors and to determine whether policy has been successful in improving the spread over time. In this paper, we construct a multidimensional composite wellbeing measure, at a micro level, which includes “economic and non-economic” and “objective and subjective measures” of wellbeing. We use NIDS data spanning the period 2008 – 2015. We compare the results on measuring wellbeing inequality using the composite index and income. This allows us to gain insight into, which is a better measure of wellbeing inequality. Additionally, we investigate the determinants of cross-sectional wellbeing inequality in 2008 and 2015 using regressions of the recentered influence function. Lastly, we use the Oaxaca-Blinder decomposition to identify the role played by covariates in shaping the evolution in wellbeing inequality. This allows us to determine if the observed change in wellbeing inequality is mainly due to a coefficient – or endowment effect. We focus on South Africa, as it is one of the most unequal societies in the world. Our results show a more equal distribution in multi-dimensional wellbeing than either income per person or life satisfaction and indicate that the spread in wellbeing has improved from 2008 to 2015. Factors that decrease wellbeing inequality are mostly demographic in nature. Factors that increase wellbeing inequality are (i) the gap in wellbeing between rural and urban areas (ii) the limited number of people that have access to computer literacy, credit and transport and (iii) relative income. Lastly, we find that improved wellbeing inequality is due to better efficiencies in the use of endowments, rather than increases in the endowments itself. Policies should not only endeavour to increase endowments but also to improve the efficient translation of these endowments into higher levels of wellbeing equality by improving institutions and limiting corruption.

Working paper 793
1 September 2019