Backfill out-of-sample populace estimates for latest release
#2
by nikhil-woodruff - opened
The latest US release was built with --skip-out-of-sample-reforms, so every out-of-sample (OBBBA / tax-expenditure) row in reform_validation.json had a null budget effect. This recomputes them by running the producer's own reform-validation simulation on the released populace_us_2024.h5 (policyengine-us 1.729.0, pe-core 3.26.11 — matching the build), and bumps the manifest sha to match.
The dashboard reads main, so this surfaces the numbers; the release tag keeps the original artifact for provenance. Recurrence is prevented by PolicyEngine/populace#175.
Withdrawing — to be filed under the correct account / in GitHub.
nikhil-woodruff changed pull request status to closed