The Globe as a Network: Geography and the Origins of the World Income Distribution

Résumé

In this paper I propose a model of the evolution of the world distribution of population and productivity from 1000 years ago until today. Taking agricultural potential and basic geographic features as inputs, this model makes predictions for population, sector shares of employment and output, knowledge creation and diffusion, trade flows, and migration flows across space for a series of points in time. I analyze these predictions through the lens of network theory, showing that population and economic activity agglomerate in locations central relative to important resources, which may be first-nature agricultural resources or second-nature stocks of expertise. I calibrate the model to population and urbanization levels at a 1 degree resolution in 1000 CE and simulate forward at a quarter century frequency, imposing a fixed relationship between fertility and the agricultural share of employment as well as a time series for transport cost parameters consistent with the historical record. This allows me to quantify the importance of first-nature geography relative to exogenous culture, institutions, or other unobserved factors; to measure the relative importance of agriculture-enhancing and trade-enhancing aspects of geography at different points in time; and to estimate an upper bound on the extent of future convergence in the world income distribution.