The US government invests over $130 billion annually in scientific research and development (R&D), with the mission of providing for the national defense, promoting social welfare, and securing economic growth and competitiveness both today and for future generations. Vannevar Bush argued in 1945 that basic science research, “performed without thought of practical ends,” would in some fraction of cases result in “important and highly useful discoveries,” but that these cases cannot be predicted in advance. And yet, the majority of federal R&D funding is targeted towards specific aims, with the assumption that increasing government expenditures for R&D will proportionally increase scientific advances in the targeted area. Since the seminal articles by Nelson (1959) and Arrow (1962), many have argued that private markets will not allocate sufficient resources to scientific R&D because the social value of scientific R&D exceeds the returns appropriable by private agents or firms. Private firms may also under-invest relative to socially optimal levels because scientific R&D is inherently an uncertain enterprise – that is, the payoff for any given R&D investment is rarely known with any certainty – and because any one firm may have insufficient scale or scope for risk-spreading. Market failures like these have historically served as the basis of arguments for government intervention in support of R&D, especially in periods of economic downturn when the political climate favors fiscal restraint. However, government interventions to promote R&D may also be imperfect or have unintended consequences – for example, crowding out of private investment – and there exists little guidance on the extent of intervention necessary. More recent work also recognizes that spillover effects across research projects – and even across scientific fields – are predictive of potential economic benefits (David et al. 1992), and furthermore marginal productivity may decrease with increasing project size (Wadman 2010). This panel brings together three papers that evaluate scientific advances arising from federal investments in R&D, via publication-related outcomes. All three employ bibliometric methods, each with a novel twist. Melkers and Welch combine bibliometrics and social network analysis to investigate how the federal EPSCoR program—designed to increase historically less-funded states’ R&D capacity—contributed to evolution of collaborative research networks, and how those networks, in turn, stimulated publication output. Sampat explores the unintended outcomes of NIH-funded research, combining data on the original disease targets of NIH awards with the subsequent patents and approved new molecular entities attributed to those funded research efforts, measuring the extent of serendipitous cross-disease research advances. Finally, Blume-Kohout, Krestel, and Smyth use topic modeling algorithms to assess whether National Cancer Institute funding increased the volume and quality of cancer-related published research, and the extent to which public funding targets riskier investments. Taken together, this panel will demonstrate a sample of our field’s progression and maturation beyond single-discipline approaches and production-function formulae, to address the policy-relevant question: how do socially desirable scientific outcomes actually arise from federal investments in R&D?