Measuring the commercial potential of science

Abstract

Abstract Research Summary We develop an ex ante measure of commercial potential of science, an otherwise unobservable variable driving the performance of innovation‐intensive firms. To do so, we rely on large language models and neural networks to predict whether scientific articles will influence firms' use of science. Incorporating time‐varying models and the quantification of uncertainty, the measure is validated through both traditional methods and out‐of‐sample exercises, leveraging a major university's technology transfer data. To illustrate the methodological contributions of our measure, we apply it to examining the impact of university reputation and university privatization of science, finding that firms' reliance on reputation may lead to foregone opportunities, and privatization (i.e., patenting) appears to increase firms' use of the science of one university. We make our measure and method available to researchers. Managerial Summary Using machine learning, we develop a measure that estimates the probability that a scientific discovery will contribute to a commercially valuable innovation. This work addresses a key challenge: the inability to observe what scientific discoveries are worth pushing forward into commercial application. We illustrate the usefulness of this measure with two examples: 1.) firms’ use of research from prestigious universities over equally promising work from less prominent ones; and 2.) how patenting affects the diffusion of commercially relevant science across firms. For practitioners, this measure can inform R&D, licensing, and other innovation related decisions by guiding firms’ search for commercially relevant scientific research. The measure and the associated code are publicly available.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 15 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.