Working Paper: CEPR ID: DP13211
Authors: Samuel Bowles; Wendy Carlin
Abstract: We represent a population as a complete undirected network, the edges of which are the fundamental data on experienced disparities. This yields a Gini coefficient (for wealth, say) for finite populations that is based on the mean wealth difference between all pairs of individuals relative to the mean wealth, which we demonstrate is not the case for the conventional Lorenz curve representation and the algorithm widely-used to calculate it. Our method also provides simple and intuitive explanations of the effects on the Gini coefficient of changes in the wage share, the employment rate, and other macroeconomic and demographic variables.
Keywords: inequality; Gini coefficient; relative mean difference; Lorenz curve
JEL Codes: D31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| network representation of wealth disparities (D31) | Gini coefficient (D31) |
| macroeconomic factors (E66) | Gini coefficient (D31) |
| wage share (D33) | Gini coefficient (D31) |
| employment rate (J68) | Gini coefficient (D31) |