Data audit — T1-forecast
All SHA-256 hashes below were recomputed and match data/README.md and bundle.json. Row counts = data rows after header.
| file | SHA-256 | rows | cols | first | last | missing values | duplicate dates | gaps |
|---|---|---|---|---|---|---|---|---|
| vintages/PAYEMS_vintage_2025-10-31.csv | d4b4c909ed321fbf8021c385beb1512ac000009e22bb20bc27c791a248d33c60 | 1040 | 2 | 1939-01-01 | 2025-08-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2025-11-30.csv | a05ef7fc50cfc339207788f9d4205a619717bc44a335488d7790ae823bdb7b77 | 1041 | 2 | 1939-01-01 | 2025-09-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2025-12-31.csv | 829dae087af61916e87994220a3e5ebf5853fa43cc163b3798715603ec8bb031 | 1043 | 2 | 1939-01-01 | 2025-11-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-01-31.csv | 8fd01c5d2740e869458721a66b81bb2e247085e1e60cb93f3cb06bedf01885d7 | 1044 | 2 | 1939-01-01 | 2025-12-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-02-28.csv | 4a913e67ac5ffbb4d127f03f6f10ce0647998779629579617dabf1344d072db3 | 1045 | 2 | 1939-01-01 | 2026-01-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-03-31.csv | 919de86b07be90466e19844b54482ef8a3f6c8da7f9fe2a3ef3af0d02923c92e | 1046 | 2 | 1939-01-01 | 2026-02-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-04-30.csv | 06a8d663d9aaa61c6d5bda15d2893b10fd127b1d64fa16d47335341f62431c7a | 1047 | 2 | 1939-01-01 | 2026-03-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-05-31.csv | e2a152e874139acdebfc1715ffff9c1e5320f639f1443fff976b389724a5932d | 1048 | 2 | 1939-01-01 | 2026-04-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-06-30.csv | 19fbc6cd3547ffad1e22f32ac76f91758859804654b77622a4f42eac2d4e286a | 1049 | 2 | 1939-01-01 | 2026-05-01 | 0 | 0 | 0 |
| vintages/PAYEMS_vintage_2026-07-31.csv | 3a3a6603fcc31c6eb2ee3293143888ab3440bbe3a962565eb39711b92019df41 | 1050 | 2 | 1939-01-01 | 2026-06-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2025-10-31.csv | 3ecb0832a391e7261145af41a7a951f823df3a08d476162721022a0922d37e36 | 404 | 2 | 1992-01-01 | 2025-08-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2025-11-30.csv | 4a683d85374690f1e73d66c4f890156d0d32cd8b27d5d67b72582b9a0dd825a4 | 405 | 2 | 1992-01-01 | 2025-09-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2025-12-31.csv | 436df4b5509b1dfe61ea11b8f98dce27f012ff6b0255eb194eea564ad7e880dd | 406 | 2 | 1992-01-01 | 2025-10-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-01-31.csv | b92ab6a5c1dbdc5653e85c56fdb7dc0646f1fca7227ed6e64f1424e17446aa9e | 407 | 2 | 1992-01-01 | 2025-11-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-02-28.csv | 2af0a3aaeeba81efddb2613d326e3f96b5a1c88c41eb30ad384ab3cc4b289a77 | 408 | 2 | 1992-01-01 | 2025-12-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-03-31.csv | 0316fdf238d55d9f109b0be85e8c2d2eb00aee57016c0328d39405b349a8815f | 409 | 2 | 1992-01-01 | 2026-01-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-04-30.csv | 02dcc697a1db34753194974b1481dbd6a3ffcfe333864850e5ddd0e3202d2b8f | 411 | 2 | 1992-01-01 | 2026-03-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-05-31.csv | 8c3fe3530213589635ad6bb086f78ed613c459bf435c4558a5c51c597e19d87b | 412 | 2 | 1992-01-01 | 2026-04-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-06-30.csv | 2f103157f4a9b7a3c90c63af2033b9570d5e334205878714bbd35c8b304d7524 | 413 | 2 | 1992-01-01 | 2026-05-01 | 0 | 0 | 0 |
| vintages/RSAFS_vintage_2026-07-31.csv | 52c04a8fce2effa727f3ef267eaee04e13aef20f8aaabc982277d08fc5b8e5b2 | 414 | 2 | 1992-01-01 | 2026-06-01 | 0 | 0 | 0 |
| RSAFS_current.csv | 627d8bb386ac80f4aede3a9496b63040df584875fb766d455b08ebd84311468f | 417 | 2 | 1992-01-01 | 2026-09-01 | 0 | 0 | 0 |
| UMCSENT.csv | 3076277f9f21f4bd3dc77fd68add84d417e5cce23f3adb96dc70313b3ad2dd46 | 887 | 2 | 1952-11-01 | 2026-09-01 | 210 | 0 | 0 |
| first_release.csv | 1d1ae8b3cacf7799aaefc7e0a969b3886e7979c36fce02ccf77d7c79fb7343e4 | 28 | 5 | 2024-06-01 | 2026-09-01 | 0 | 0 | 0 |
Schema and units
- RSAFS vintages / current:
observation_date(YYYY-MM-01), level in millions of USD, SA. Integers, no missing values, no duplicate dates, no internal gaps. - PAYEMS vintages: thousands of persons, SA, 1939-01 onward, complete.
- UMCSENT: index 1966Q1=100; 210 empty fields, all before 1978 (quarterly survey period; last empty 1977-12). Complete from 1978-01 to 2026-09.
- first_release.csv: 28 target months 2024-06 … 2026-09; recomputing 100·ln(level_first_release/level_prev_year_same_vintage) reproduces
yoy_first_releaseto 6 decimals; for the 10 rows whose first_release_vintage is one of the shipped vintages, the levels match the shipped vintage files exactly.
Real-time design: last(V) per origin (RSAFS)
| origin V | last(V) | gap (months before V) | h=1 target | h=3 target |
|---|---|---|---|---|
| 2025-10-31 | 2025-08 | 2 (delayed) | 2025-09 | 2025-11 |
| 2025-11-30 | 2025-09 | 2 (delayed) | 2025-10 | 2025-12 |
| 2025-12-31 | 2025-10 | 2 (delayed) | 2025-11 | 2026-01 |
| 2026-01-31 | 2025-11 | 2 (delayed) | 2025-12 | 2026-02 |
| 2026-02-28 | 2025-12 | 2 (delayed) | 2026-01 | 2026-03 |
| 2026-03-31 | 2026-01 | 2 (delayed) | 2026-02 | 2026-04 |
| 2026-04-30 | 2026-03 | 1 (two months added at once) | 2026-04 | 2026-06 |
| 2026-05-31 | 2026-04 | 1 | 2026-05 | 2026-07 |
| 2026-06-30 | 2026-05 | 1 | 2026-06 | 2026-08 |
| 2026-07-31 | 2026-06 | 1 | 2026-07 | 2026-09 |
All 20 target months are present in first_release.csv. Because of the publication delays the target months are not evenly spaced in calendar time (h=3 targets skip 2026-05; h=1 targets skip 2026-03).
Anomalies / issues
- Irregular publication timing (2025 federal shutdown period): vintages 2025-10-31 … 2026-03-31 end two months before the vintage date; 2026-04-30 adds 2026-02 and 2026-03 at once. Handled by defining targets relative to last(V), as TASK.md prescribes. The first-release dates of 2025-09 (2025-11-30 vintage) and 2026-02 (2026-04-30 vintage, same as 2026-03) reflect this.
- Revisions inside the vintage grid: the last ~3 months are revised at each vintage (e.g. 2026-05 level 763705 → 766876). Real-time forecasts use the vintage values as published; actuals are first releases.
- Large level shift (benchmark/annual revision) in later data: RSAFS_current levels are ~3.5–4% below the 2026-07-31 vintage for all months shown from 2024-06 on; first_release.csv shows the shift between the 2026-08-31 and 2026-09-30 vintages (2026-07 first-release level 763602 vs current 729915; 2026-08 base-year level 700218 vs 731700 in shipped vintages). YoY growth is computed within one vintage, so the level shift largely cancels, but it inflates estimand (d) and means the h=3 actuals for the last two origins (2026-08, 2026-09) come from post-revision vintages. Documented, not corrected (the task defines the actual).
- PAYEMS publication was also delayed (vintage 2025-10-31 ends 2025-08; 2025-12-31 adds 2025-10 and 2025-11). Not used by my model.
- UMCSENT is a current-vintage file; per README it is not revised, so restricting to months ≤ month(V) is real-time. Not used by my model.
Availability-rule violations in the inputs
RSAFS_current.csvcontains post-origin information; it is read only for estimand (d).first_release.csvcontains post-origin information; used only as the evaluation actual.- No other violation. No rows are excluded (see out/exclusions.csv).