Private investment in innovation is often at sub-optimal levels because of various types of market failure. In response, governments support private innovation initiatives by lowering their cost through, for instance, tax incentives or subsidies for R&D projects. In the last years, governments have increasingly aimed at not only boosting but also influencing the direction of R&D investment via direct procurement. Governments may directly acquire R&D-based goods and services, and this activity can also have indirect effects on firms’ innovative output and, ultimately, productivity. This project has analyzed the determinants of the efficacy of such a policy. First, we studied the implications of bottlenecks within US federal procuring offices for the pursuing their mission effectively. We merged for the first time different data sources covering every step of the acquisition process with the identity of the bureaucrat in charge of the purchase. We found that an increasing workload burden to the same officer, measured in terms of employment shortage in their agency - holding the budget and the number of purchases constant - impacts negatively the probability of R&D contracts awarded by him/her to generate future patents. Second, we have studied the indirect effects of procurement on firm-level productivity. This analysis has been realized by exploiting public demand shocks to firms (i.e., the unexpected award of a procurement contract). These shocks possibly induce distortions in market structure and affect firms’ incentives to innovate. We set up a novel and rich dataset of Italian procurement auctions that includes losing participants and their bids. We found that these “procurement firms” survive more in the market without a concurrent increase in their productivity. They become increasingly dependent on the public source of revenues and disincentivized to invest in innovation as immune from market dynamics.