Recent empirical research has documented the importance of shocks to firm- specific productivity, but has provided only limited evidence on their sources. This paper proposes and analyzes purposeful experimentation by firms as a source of such shocks and models industry dynamics in such a setting. We thereby make two contributions. The first is conceptual and consists in providing a microfoundation to the stochastic process for firm-level productivity typically specified in the macroeconomic literature with firm heterogeneity. The second consists in quantifying the importance of experimentation for aggregate productivity growth to which experimentation, as a generalized form of R&D, contributes. We show that in a setting that allows for growth through experimentation, through market selection among firms and because of other sources, 45% of aggregate productivity growth can be attributed to experimentation. As a consequence, allocative distortions may not just reduce the level of productivity, but also have a substantial effect on growth rates.
