Why Has the Number of Billionaires Increased So Much?

We study the fourfold increase in the number of billionaires since 2001, and its regional variation. We develop a model where wealth is proportional to the length of a Self-Avoiding Walk on a random network, which rationalizes the Gompertz distribution of log wealth in our data. The model predicts the elasticity of top inequality to depend solely on population size and the lower thresholds for wealth to depend solely and one-for-one on regional GDP per capita and a global asset-market factor. All predictions hold in our data. Time fixed effects do not significantly improve the model fit, but regional effects do. Counterfactual exercises closely predict observed mean (log) wealth and billionaire numbers. The increases in billionaire numbers and mean (log) wealth are almost entirely driven by increases in GDP per capita. We interpret our results in the context of Melitz (2003)-style models where market size shapes firm (and hence wealth) concentration.