Replication Package for Oil Price Innovations, Shock Heterogeneity, and the Limits of Financial Amplification

This record contains replication round 6 for the manuscript “Oil Price Innovations, Shock Heterogeneity, and the Limits of Financial Amplification.” The archive includes the clean and executed notebooks, executable Python modules, cached public-source data, intermediate audit tables, machine-readable results, journal-ready figures, documentation, and integrity checks required to reproduce the empirical analysis. The workflow reproduces the baseline and financial VAR models, asymmetric WTI specifications, 120-month rolling VAR estimates, subsample comparisons, HAC local projections, and the recursively identified structural oil-market shocks. The structural analysis distinguishes adverse oil-supply, aggregate-demand, and oil-specific-demand shocks. The common transformed sample covers February 1990–January 2026 and contains 432 monthly observations. The VAR(2) models use 430 effective observations. The cached data permit offline replication and no API key or other credential is required. Data sources include Federal Reserve Economic Data, the U.S. Energy Information Administration, and the Federal Reserve Bank of Dallas. Bundled source data retain the terms and provenance of their original providers. To reproduce the analysis, extract the archive and run: python -m pip install -r requirements.txtpython run_all.py Further details are provided in README.md, LICENSE_NOTES.md, REPLICATION_MANIFEST.json, and CHECKSUMS.sha256.