# Stage 131 — official macroeconomic data ingestion and provenance

The ingestion pipeline requests 2015–2024 annual observations for Israel, USA, Germany, France and UK from World Bank WDI: nominal GDP in local currency (NY.GDP.MKTP.CN), employment-to-population ratio (SL.EMP.TOTL.SP.ZS), GDP per person employed in constant purchasing-power dollars (SL.GDP.PCAP.EM.KD), and CPI (FP.CPI.TOTL). These series are **not directly interchangeable**: GDP and CPI levels have different units, employment rate is not unemployment, and GDP per employed person is only a proxy for labor productivity.

Wages are explicitly marked NOT_SOURCED until a suitable OECD/ILO wage measure is verified. The program validates country, indicator, year, finite values and basic bounds; it preserves missing years and individual source URLs. Network failures are recorded, not converted to observations. No parameter in the macroeconomic model is calibrated or modified.

Run: `python scripts/official_macro_pipeline_v131.py --output data/processed/official_macro_pipeline_v131.json`. Results must be reviewed for source quality, revision vintage, coverage and cross-country comparability before estimation.

## Verified ingestion — GitHub Actions run 423

All 20 WDI country-indicator requests reported status `fetched`, with zero fetch failures and **200 observed_api records** (5 countries × 4 indicators × 10 years, 2015–2024). The output was retained as the CI artifact `official-macro-pipeline-v131`. This establishes API coverage for the requested cells, not source harmonization or structural identification.

Wages remain **NOT_SOURCED**. The GDP per person employed indicator is a proxy and cannot be treated as a causal automation productivity estimate. The employment-to-population ratio is not the unemployment rate. CPI levels and GDP levels should not be compared directly across countries without appropriate normalization and units. No structural parameters were updated. The data are subject to subsequent WDI revisions; retrieval date and source vintage should be recorded before estimation.