This paper develops and analyzes a macroeconomic model in which aggregate growth and fluctuations arise from the discovery and diffusion of new technologies; there are no exogenous aggregate shocks. The temporal behavior of aggregates is driven by individuals' efforts to innovate and/or make use of others' innovations. Parameters describing preferences, production possibilities and learning technologies are estimated using post-war U.S. data. The model delivers predicted aggregates that grow and fluctuate much like the data. The key features of post-war growth are explained by new technologies that differ in terms of the magnitude of their improvement over existing methods and the difficulty of acquiring them. The model implies a negative trend in technological dispersion, and that the generally lower growth witnessed during the last two decades is the result of new technologies offering comparatively minor or less broadly-applicable improvements. Data on the growing and fluctuating share of engineering Ph.D.s support the model's technological interpretation of the growth facts, and data on patent applications and adult schooling are consistent with the notion that newer technologies are more specific and proprietary.