Firm-year analysis datasets¶
These products aggregate model scores or assemble outcomes at firm-year grain.
The Compustat-CRSP MQ dataset is the current path for new analysis work: it uses annual Compustat as its spine and attaches RCID-year match quality, CRSP, and patent data in one place.
flowchart TB
rcid_mq[(RCID-year MQ)]
rgk[(Revelio-Compustat mapping, rcid-gvkey)]
comp[(Processed annual Compustat)]
ccm[(Project-raw CCM link table)]
volatility[(Processed CRSP)]
patent_panel[(Patent firm-year data)]
mq_dataset[(Compustat-CRSP MQ dataset)]
rcid_mq --> mq_dataset
rgk --> mq_dataset
comp --> mq_dataset
ccm --> mq_dataset
volatility --> mq_dataset
patent_panel --> mq_dataset
classDef input fill:#e8eef7,stroke:#52739e,stroke-width:2px
classDef analysis fill:#eee5f5,stroke:#76528c,stroke-width:2px
class rcid_mq,rgk,comp,ccm,volatility,patent_panel input
class mq_dataset analysis
Datasets and roles¶
| Dataset | Role | Grain |
|---|---|---|
| RCID-year MQ | Universal reusable score aggregation | Revelio employer-year |
| RCID-occupation-year MQ | Universal reusable score aggregation by job category | Revelio employer-occupation-year |
| Company-year MQ | Reusable across analysis-data pipelines | Company-year |
| Patent firm-year data | Patent outcomes | GVKEY-year |
| Processed CRSP | Return volatility and compounded return | Security-month |
| Compustat-CRSP MQ dataset | Direct analysis input; built after RCID-year MQ | Compustat GVKEY-fiscal-year, plus sample status for training profiles |
rnd_mq_panel.parquet |
Configured but disconnected local support | Company-year |
Company-year MQ
company_year_mq.parquet aggregates the worker-firm-period scores by
company and year. It does not feed the firm-performance or innovation
tables, but remains in use by the CHRO and deals branches.
The Compustat-CRSP MQ dataset page documents that dataset's inputs and design in more detail.