Research into the sources of productivity growth has taken many disparate routes, from theoretical modeling using an aggregate perspective, to empirical explorations using firm-level data. This paper shows that models relying on the construct of the representative firm do not provide much scope to consider the effects of policy on productivity. The paper provides examples from the literature of the evidence against the representative firm assumption. Once the model is extended to include entry, exit, and heterogeneous firms, policy is able to affect aggregate productivity through many paths, including the efficiency of resource allocation and the margin of selection. As an example, some empirical explorations are provided for the effects of idiosyncratic distortions to firm profitability, the effects of exit costs, and the types of policy that may boost intangible investments.1 I am very indebted to my co-authors from the research projects cited in this paper: Pieter Gautier, John Halitwanger, Enrico Perotti, Stefano Scarpetta, and Joris de Wind, and all participants of the ONS/Eurostat ICT Impacts project.