Welcome to the ACHP Data Library
Data is provided by: Utku Acikalin (Cornell University)
Tolga Caskurlu (University of Amsterdam)
Embedded Files
Patent-Level Alice Exposure Scores (w/ sample flags and case-overlap indicators):
Our Alice exposure scores are estimated using a fine-tuned ModernBERT large language model trained on patent text and USPTO rejection records. The paper itself uses firm-level treatment variables that aggregate these patent-level scores into a single value per firm via Equation 2 of the paper, whereas the data here is more detailed and provides the underlying patent-level scores from which any firm-, industry-, or portfolio-level measure can be constructed. The data thus contains one row per pre-Alice granted U.S. utility patent in CPC categories affected by the Alice decision and is a more granular data structure than firm-level aggregates. Each patent also associates with two flags indicating whether the patent is potentially affected by the related Bilski, Mayo, or Myriad Supreme Court decisions, and whether the patent is included in the reduced sample used in the paper's baseline regressions. Please read the readme file and the foundational paper used to create these measures noted below.
Download Patent-Level Alice Exposure Scores. [Download Alice Scores] [View Readme]
The following study provided the key innovations to the creation of this data:
Intellectual Property Protection Lost and Competition: An Examination Using Large Language Models — Utku U. Acikalin, Tolga Caskurlu, Gerard Hoberg, and Gordon M. Phillips, forthcoming at Journal of Financial Economics.
Page updated
Google Sites
Report abuse