Abstract
A growing body of evidence documents a long-run decline in the disruptiveness and novelty of scientific output, despite rising research investment and a fast-growing scientific workforce. The dominant interpretation attributes this slowdown to an exogenous "burden of knowledge", breakthrough ideas are becoming harder to find as low-hanging fruit is exhausted.
This paper proposes and tests an alternative, endogenous explanation: the rise in competition for funding, positions, and publications within academia has progressively shifted scientists' idea selection toward safer, more conventional projects, narrowing the set of trajectories actively explored.
Using the American Recovery and Reinvestment Act (ARRA) of 2009 as an exogenous, one-time positive shock to NIH funding availability, which varied substantially across NIH Institutes and sub-fields, I apply a continuous difference-in-differences design to test whether the reduction in field-level competition induced by ARRA translates into a measurable post-shock increase in the novelty of publications.
Research Question
Has the long-run intensification of competition in U.S. biomedical science contributed causally to the decline in the novelty and disruptiveness of publicly funded research?
Key Contributions
Methodology
Scientific novelty measured along three independent dimensions: MeSH-based novelty (new term introductions), text-based novelty (novel noun-phrase combinations in titles/abstracts), and disruptiveness via the CD-index.
NIH application success rates, scientists-per-funding-dollar, share of papers in top-decile journals relative to field size, and share of non-tenure-track positions.
The 2009 ARRA shock to NIH funding is exploited as a quasi-experiment, with treatment intensity varying continuously across NIH Institutes and sub-fields. Effects are expected to emerge at the typical 2–5 year publication lag.
Interactive data
Explore the underlying grant landscape directly. The interactive explorer covers more than one million NIH grants across ten institutes, filterable by activity type, application type, institute, study section, and ARRA status.
Keywords
Interested in the full working paper, data, or methodology? Send a message and I'll get back to you directly.