Medical Research as a Productivity Indicator

Across fields, the quantity of research has increased substantially, without an attendant increase in output. We argue that, in medicine, this indicator of declining productivity reflects a compositional shift toward low-capital, low-productivity research. Using a fine-tuned, open-source large language model, we construct a novel census of capital-intensive, high-productivity medical investment -- clinical trials. Since 2010, the annual quantity of clinical trials has been constant. By contrast, the quantity of other forms of clinical research has increased substantially. Within clinical trials, there is substantial heterogeneity in productivity. The distribution of this heterogeneity is stable over time.