Impute CPS ORG variables (hourly wages, union status) onto eCPS #587
Description
Summary
Impute CPS Outgoing Rotation Group (ORG) variables onto the eCPS to enable wage-rate and labor market analysis. The ORG contains several variables not available in the ASEC that are important for policy modeling.
Variables to impute
These are ORG-unique — not available in the ASEC:
| Variable | Description | Policy relevance |
|---|---|---|
| Hourly wage (actual reported) | Direct hourly pay for hourly workers, usual weekly earnings for salaried | Minimum wage, wage subsidies, overtime policy |
| Paid-by-the-hour flag | Whether worker is paid hourly vs salaried | Overtime eligibility, wage floor analysis |
| Union membership/coverage | Union member or covered by union contract | Labor policy modeling |
| Overtime/tips/commissions | Weekly OTC amount (combined field) | Tip credit policy, overtime rules |
Motivation
- Hourly wages are the gold standard for wage analysis (EPI, CEPR, BLS all use ORG for this). Deriving hourly wages from ASEC annual earnings ÷ hours is noisy and biased.
- Union status is only collected in the ORG — there is no ASEC equivalent.
- External users are already building workarounds: eig-wagesubsidy-policy-sim uses ORG separately because the eCPS lacks these variables, forcing a 4-household-type approximation instead of using PolicyEngine's Microsimulation class.
Approach
Use the existing QRF (microimpute) pipeline, same pattern as SIPP/ACS/SCF imputation in source_impute.py:
- Load ORG microdata (available from EPI, CEPR, or IPUMS)
- Train QRF on shared covariates: employment_income, age, education, occupation, industry, state, sex, race, hours worked
- Impute onto eCPS ASEC records
- Validate imputed distributions against ORG originals by subgroup
Data access
ORG is public use data, no licensing restrictions. EPI provides cleaned extracts with constructed hourly wages (Pareto top-code adjustment, allocation flag filtering).