Knowledge diffusion, rising inequality, and the dynamics of a pulled front

Abstract A model of economic growth is presented in which workers increase their labour income by learning from higher-paid workers. They also experience individual random income shocks. A calibration of the model to U.S. wage micro data implies that these shocks are Brownian ‘noise’, with drift close to zero. Growth comes from occasional random advances at the frontier, which create novel learning opportunities for others and drive sustained growth. The long-run growth rate is increasing in both the variance of shocks and the knowledge-diffusion rate. The model closely reproduces the evolving shape of the U.S. wage distribution between 1991 and 2023.