Stimulating Peer Effects? Evidence from a Research Cluster Policy

Production of knowledge relies on peer effects and interactions between researchers. However, little is known on how much policies may stimulate these peer effects. In this paper we shed light on this question, and show how a public "research cluster'' policy, which funds local networks of researchers working on a common theme, affects the organization of research within these clusters and the productivity of its members. Using data from a large scale financing program in France, and relying on an identification strategy based on grades awarded by reviewers, we show that members of financed clusters increase by up to 30% the research collaborations they have with other members of the cluster, compared to researchers of non selected proposals. This very large reorganization of the research network translates into a more modest positive effect on research productivity. Paradoxically, those who benefit the most from the financing, are those who were not at the core of the research topic, i.e. were not cited in the bibliography of the research proposal, who significantly increase their links with core members and their total publication counts. Consistently, the policy reduces inequality in publication outcomes within the cluster. It stimulates peer effects to the benefit of periphery members