Patent data

The repository contains a Python pipeline that aggregates KPSS patents to CRSP-security years and attaches historical CRSP industry fields.

flowchart TB
    patents[(KPSS patents)]
    patent_links[(KPSS patent-PERMNO matches)]
    crsp_names[(CRSP historical names)]
    panel[(Processed patent firm-year data)]

    patents --> panel
    patent_links --> panel
    crsp_names --> panel

    classDef source fill:#e8eef7,stroke:#52739e,stroke-width:2px
    classDef reusable fill:#e2f2e7,stroke:#3b7a4b,stroke-width:2px
    class patents,patent_links,crsp_names source
    class panel reusable

Current status: no downstream consumers

The processed patent firm-year dataset is not read by the current firm-performance or innovation pipeline. The active R analysis builder reads raw KPSS_2024.csv directly and writes a separate analysis patent panel.

Both pipelines are documented because they implement different transformations and both outputs currently exist in the project design. Building the Python product is not a prerequisite for the patent regressions.