# Data audit (P1)

| file | SHA-256 | rows (excl. header) | cols |
|---|---|---|---|
| data/cpi_quarterly.csv | c8652dec1567ec35e3b84ee8fa55545e9bf0ae4990a4d6d05b89a6d22f6801ea | 127 | 3 (quarter, cpi_avg, infl_annualised) |
| data/cpi_monthly.csv | 855d069176369d30d02cbbe0cc6456b76d2f3967cdaabc1b362c2461ef9015a2 | 381 | 2 (month, cpi) |
| data/README.md | 8ddc03b23fc4efce8ce697879fbb2ddca04c40ae54eb772a04e9777444098d0a | — | — |

All hashes match README.md and bundle.json.

- Units: cpi index (2002=100, NSA); infl_annualised in percent at annual rate.
- Date ranges: quarterly 1995-01-01 … 2026-07-01 (127 consecutive quarters, all start months in {1,4,7,10}); monthly 1995-01 … 2026-09 (381 consecutive months).
- Missing: only infl_annualised for 1995Q1 (by construction, no lag). No duplicates in either file.
- Derivation check: quarterly mean of monthly values reproduces cpi_avg (max abs diff 3.3e-7, every quarter has 3 months); 400·Δln cpi_avg reproduces infl_annualised (max abs diff 3.3e-6, rounding).
- Distribution: mean 2.13, sd 2.39, min −6.04, max 12.15 (2022Q2 — genuine inflation surge, not an error; kept). Strong seasonality: mean by quarter Q1 2.32, Q2 4.18, Q3 1.66, Q4 0.30.
- Availability: last target needed is 2026-07-01 (origin 2025-07-01, h=4) — present. No input violates the availability rules; the analysis code truncates the series at each origin before any estimation.
- Exclusions: 1995Q1 dropped (no inflation value). See out/exclusions.csv.