Stage 130 — empirical calibration source-quality gate

The first empirical calibration work package checks nine rounded Israeli corporate tax receipts/GDP observations from mixed OECD editions. It reports descriptive levels and leave-one-observation-out sensitivity, preserving source URLs and publication vintages.

This is not a calibrated macroeconomic model. Tax revenue as a share of GDP cannot identify an effective corporate tax rate, tax base, productivity elasticity, price pass-through or redistribution capacity. No structural parameters are updated. Missing years and mixed editions prevent treating the observations as a harmonized time series.

The next empirical work requires verified, aligned official annual GDP, tax-base, employment, price and productivity series, plus a prespecified identification strategy. Do not fill gaps by interpolation and present them as observed.

Reproduce: python scripts/empirical_calibration_gate_v130.py --output data/processed/empirical_calibration_gate_v130.json.

Verified findings — GitHub Actions run 413

Nine rounded, sourced Israeli corporate tax receipts/GDP observations were accepted. Descriptive mean: 3.233333%; median 3.2%; minimum 2.6%; maximum 4.3%. The leave-one-observation-out mean ranges from 3.1% to 3.3125%.

These are descriptive statistics for a sparse, mixed-vintage dataset, not calibrated structural parameters. The validation gate explicitly reports calibrated=false. No simulator parameters were modified. An independent official series and a defensible identification strategy remain prerequisites for estimating the effects of automation on labor, prices, corporate tax bases and transfers.