Recent evidence suggests significant changes over time in the pattern of comparative advantage across countries and industries. What drives such dynamics remains an open question. A mechanism suggested by theoretical
literature, but not yet brought to bear on this evidence, is learning-by-doing. In this paper I develop a quantitative model of trade and growth with the goal of characterizing the relationship between learning-by-doing and the dynamics of comparative advantage.
The model features an occupational dimension to learning, which endogenously generates a particular network structure of inter-industry learning diffusion, based on occupational similarity. The model predicts that countries with a comparative advantage in
industries more central in this network will grow more in the aggregate. I use the model-implied dynamics of comparative advantage to quantitatively discipline the amount of occupational learning and the extent to which learning diffuses across industries.
Compared to intra-industry learning, I find that cross-industry learning diffusion explains at least four times as much of the dynamics of comparative advantage, as well as forty percent of an industry's contribution to aggregate growth.