Skip to content

Executive compensation

This analysis tests whether executives' MQ is associated with their total compensation and salary after linking ExecuComp records to Revelio workers.

flowchart LR
    data[(Training and full-training<br/>ExecuComp regression data)]
    prepare[[Select three samples and<br/>standardize MQ within outcome]]
    fit[[Estimate compensation ladders]]
    latex[/execucomp_regressions.tex/]

    data --> prepare --> fit --> latex

    classDef input fill:#eee5f5,stroke:#76528c,stroke-width:2px
    classDef code fill:#f3f3f3,stroke:#666,stroke-width:2px
    classDef tableOutput fill:#e2f2e7,stroke:#3b7a4b,stroke-width:2px
    class data input
    class prepare,fit code
    class latex tableOutput
Inputs
Dataset Location
ExecuComp regression data /data/<coverage>/analyses/<mode>/execucomp_regression_table.parquet
Build
Component Location
Builder analyses/modules/disaggregated_execucomp.R
Runtime hct-tower-analyses-cpu
Entry point analyses/pipelines/disaggregated_execucomp.R / build-execucomp-regression-table, execucomp-regressions
Output
Dataset File / location Grain and variables
Executive-compensation results execucomp_regressions.tex in /artifacts/<coverage>/analyses/disaggregated_execucomp/ LaTex table for outcome, sample, and specification combinations

The table covers training out-of-sample, training in-sample, and full-training in-sample observations for log total compensation and log salary. Each outcome and sample uses five specifications: MQ only; accounting controls; and the controlled model with year, firm, or firm-plus-year fixed effects. Standard errors are clustered two ways by Revelio worker and firm.

Build both mode-specific regression datasets before estimation.

Command reference
Invocation and option Required / default Source-defined effect
build_regression_table --config PATH Required Builds the selected mode's ExecuComp regression table.
build_regression_table --mode training\|full_training full_training Selects the model score mode.
build_regression_table --overwrite Off Replaces an existing regression table.
run_regressions --config PATH Required Fits and formats the executive-compensation regressions. No additional options are accepted.
Option Required / default Source-defined effect
--task build-execucomp-regression-table Required Builds a regression table; --mode defaults to full_training and accepts training or full_training; --overwrite is forwarded.
--task execucomp-regressions Required Runs regression estimation; the wrapper rejects --mode.
--config CONFIG.yaml Required Resolves the analyses config.
--partition, --time, --mem, --cpus, --job-name, --dependency, --dry-run Wrapper defaults Standard CPU overrides. The table builder default is 8 CPUs, 64 GB, 24 hours; regression task is 8 CPUs, 64 GB, 4 hours.
cd /hct-tower/analyses
Rscript --vanilla pipelines/disaggregated_execucomp.R build_regression_table \
  --config config/config_one_percent.yaml --mode training --overwrite
Rscript --vanilla pipelines/disaggregated_execucomp.R build_regression_table \
  --config config/config_one_percent.yaml --mode full_training --overwrite
Rscript --vanilla pipelines/disaggregated_execucomp.R run_regressions \
  --config config/config_one_percent.yaml
cd /hct-tower
slurm/submit.sh --task build-execucomp-regression-table \
  --mode training --config config_full.yaml --overwrite
slurm/submit.sh --task build-execucomp-regression-table \
  --mode full_training --config config_full.yaml --overwrite
slurm/submit.sh --task execucomp-regressions \
  --config config_full.yaml

The regressions are compact and are refitted on each run. The current pipeline does not preserve fitted models, numeric CSV results, or analysis metadata.