Proponents of basic science argue that objective scientific understanding can inform improvement in public policy. We gather data on scientific research cited in official benefit-cost analyses produced by US federal regulatory agencies to justify policy decisions between 2008 and 2012. We construct a science-policy network in which benefit-cost analyses and the studies they cite are the nodes, and citations represent the edges. We assess two features of each scientific publication in the network; how frequently is it used; and how broadly it spans across the network, as measured by betweenness centrality. We ask which author affiliations and funders are associated with the best-cited and farthest spanning publications. Elite universities and major government funders support publications that are most heavily cited, but the farthest spanning articles are written by scientists with non-academic affiliations and sponsored by non-governmental funders. These results suggest that bias towards academically affiliated investigators should be scrutinized by major funding organizations if a major objective is to support science that is used by policymakers.