Key points are not available for this paper at this time.
A simple aggregate growth model is presented in which technology is described by a probability distribution from which new plants are drawn. Especially good draws are viewed as technological innovations that shift the mean of the following period's plant distribution function. The resulting technical change is endogenous, random, and cumulative. In contrast to conventional growth models, the model's growth path displays nonstationary drift rather than deterministic trend, and the long-run per capita growth rate has positive rather than zero sensitivity to the model's saving parameter.
John Conlisk (Wed,) studied this question.