View article

This paper analyzes a decentralized process for the diffusion of knowledge. In equilibrium, the economy converges from an initial distribution of knowledge over agents to the steady-state distribution, which is unique. Because of the public good aspect of information, too little learning takes place, and ideas are implemented too early. The key difference between earlier formulations of search externalities by Diamond, Mortensen, and Spence on the one hand, and our own on the other, is that here spillovers of knowledge depend not only on how hard people are trying, but also on the differences in what they know: if all of us know the same thing, we cannot learn from each other. The model also addresses the following two substantive questions: first, the relationship between inequality and growth, noted some time ago by Kuznets, and second, the effect on growth of improvements in the communication technology.