Firm-year MQ statistics

This analysis-support dataset summarizes the full distribution of scored worker-firm observations within each company-year.

flowchart LR
    scores[(Worker-firm-period scores)] --> aggregate[Aggregate MQ distribution]
    aggregate --> output[(Firm-year MQ statistics)]

    classDef source fill:#e8eef7,stroke:#52739e
    classDef process fill:#fff0cc,stroke:#a66b00
    classDef data fill:#eee5f5,stroke:#76528c
    class scores source
    class aggregate process
    class output data

Grain and key: company_id, calendar year
Location: /data/<coverage>/analyses/full_training/firm_mq_stats.parquet
Active builder: build_firm_mq_stats() in analyses/modules/build_table89_dataset.R

The output stores mq_n, mq_mean, mq_var, and mq_median. The run_build_table89_dataset.R entry point builds it only when the configured file is absent, then passes it to the final public-firm dataset builder.

Rscript pipelines/run_build_table89_dataset.R \
  --config config/config_tables89_full.yaml

This aggregation is separate from company-year MQ, which stores score-weighted mean MQ and attaches broader firm information.