Working Paper: CEPR ID: DP18634
Authors: Coen Teulings; Simon Toussaint
Abstract: We study the shape of the global wealth distribution, using the Forbes List of Billionaires. We develop simple statistics based on ratios of log moments to test the default assumption of a Pareto distribution, which is strongly rejected. Hazard rates show that the log-transformed data instead follow a Gompertz distribution, which means that the data in levels follow a truncated-Weibull distribution. We further apply our model to the U.S. city size distribution and the U.S. firm size distribution. These distributions also show a rejection of Pareto in favor of (truncated-)Weibull. We discuss some theoretical and practical implications of our results.
Keywords: wealth distribution; city and firm size distribution; tail distributions
JEL Codes: D3; E2; G5
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Global wealth distribution among billionaires does not conform to Pareto distribution (D39) | follows a truncated-Weibull distribution (C46) |
| Log-transformed data supports Gompertz distribution (C51) | indicates wealth distribution is less skewed to the right than predicted by Pareto model (D39) |
| Weibull distribution provides a better fit for predicting mean wealth across various subregions (C46) | compared to Pareto model (C52) |
| Weibull model can replace Pareto in many economic contexts (C59) | implications for future research on wealth distribution and economic modeling (D30) |