We reexamine several bodies of data on the growth of output, labor, and capital, within the context of a model that admits the possibility of an externality to the capital input. The model is an augmented version of Paul Romer's (1987) reformulation of the Solow model. Unlike Romer, however, we...

I estimate a model in which new technology entails random adjustment costs. Rapid adjustments may cause productivity slowdowns. These slowdowns last longer when retooling is costly. The model explains why growth-rate disasters are more likely than miracles, and why volatility of growth relates...