Unfair Inequality Measurement in China and South Africa

Abstract

This thesis applied two ways to measure unfair inequality in China and South Africa. Firstly, we focus on the unfair inequality violating the principle of Equal Opportunity (EOp), which answers to what extent individuals’ income difference are due to factors beyond their control (in Roemer's terminology ‘’circumstances’’). Utilizing the China Family Panel Studies Survey, we measure this Roemerian unfair inequality in individual earnings in aggregate and for each of 10 birth cohorts from 1955-1985. The aggregate result shows that Roemerian unfair inequality takes up nearly 24% of the overall inequality. The cohort pattern shows an increasing trend of Roemerian unfair inequality for the younger cohorts. Among all circumstance variables, gender is the most influential one, contributing to nearly half of the unfair inequality. But the impact of gender decreases for the later cohorts. In South Africa, we use the National Income Dynamics Study (NIDS) to measure this Roemerian unfair inequality in individual annual gross income. On average, Roemerian unfair inequality takes up nearly 16% of the overall inequality. This unfair inequality ratio shows an increasing trend for the younger cohorts. Among all circumstance variables, parental education and race are the most influential ones, both contributing to nearly 23% of the unequal opportunity. Second, we reconcile both the principle of Equal Opportunity (EOp) and Freedom from Poverty (FfP) to measure unfair inequality (Hufe, Kanbur & Peichl, 2018 )In China, HKP unfair inequality takes up 27% of the outcome inequality and remains relatively stable over the cohorts. Further decomposing HKP unfair inequality, we find that the stability results from a combination of an increasing trend of inequality violating EOp and decreasing trend of inequality violating FfP. In South Africa, the HKP unfair inequality takes up 21% of the outcome inequality. After the decomposition, we find an increasing share of inequality violating FfP for the younger cohort. Finally, we tried to include 2560 individuals with non-positive income in CFPS dataset by using a concave log-like transformation and repeat our two measurements above. With this adjustment, we find significant upward corrections of unfair inequality.