HR leadership data

This family identifies annual HR leaders at Revelio firms and creates a linked public-company CHRO and top-HR panel without using HCT scores.

flowchart TB
    positions[(Revelio positions)]
    mapping[(Revelio-Compustat mapping)]
    comp[(Processed annual Compustat)]
    ccm[(CRSP/Compustat CCM links)]
    classify[Classify titles and<br/>select annual leaders]
    leaders[(HR-leadership firm-year data)]
    public[(Public-company CHRO data)]

    positions --> classify --> leaders
    leaders --> public
    mapping --> public
    comp --> public
    ccm --> public

    classDef source fill:#e8eef7,stroke:#52739e
    classDef process fill:#fff0cc,stroke:#a66b00
    classDef data fill:#e2f2e7,stroke:#3b7a4b
    class positions,mapping,comp,ccm source
    class classify process
    class leaders,public data

Grain: one Revelio firm-year or public gvkey-year
Location: /data/full/raw/chro/

The reusable pipeline has two principal products:

Product	Builder
revelio_hr_exec_firm_year.parquet	chro_panel.py classifies titles and selects strict CHRO and top-HR leaders
revelio_chro_public_firm_year.parquet	chro_public_panel.py adds gvkey and optional permno

python -m shared.data.build_datasets.chro_panel
python -m shared.data.build_datasets.chro_public_panel

The first builder reads raw Revelio positions because it needs title fields not retained in the canonical employment-event contract. It can optionally write revelio_hr_exec_spells.parquet. The public-panel builder joins the Revelio-Compustat mapping and annual Compustat, optionally attaches a year-valid PERMNO from CCM, and writes two coverage sidecars.

ExecuComp titles provide a partial external validation through shared.data.build_datasets.validate_chro_execucomp. The separate CHRO with model MQ dataset belongs to the analysis layer.