Replication Data for: A Bayesian VAR Framework for US Macro-Financial Forecasting and Stre

Eleven US macro-financial variables from 1990 to 2024 support a Bayesian vector autoregression model for forecasting and stress-test scenario design. The deposit contains raw quarterly series from FRED and Yahoo Finance, processed estimation panels, and generated adverse scenarios. Muhammad Umar Shinwari provided the replication data and R code, last updated in July 2026.

Use Cases

• Forecasting US macro-financial conditions based on the described Bayesian VAR framework.
• Designing severely adverse stress-test scenarios based on the Federal Reserve's 2025 supervisory file.
• Conducting Growth-at-Risk analysis using the described quantile regression outputs.
• Reproducing the paper's tables and figures using the provided R pipeline and processed data.

Strengths

• Includes raw and processed data for eleven variables over a 34-year period (1990-2024).
• Provides a complete R pipeline that maps scripts to every table and figure in the associated paper.
• Contains generated adverse scenarios and the Federal Reserve's 2025 supervisory severely adverse scenario file.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.

Provenance

Source

FRED, Yahoo Finance, and the Federal Reserve's supervisory scenario file.

Collection Method

Quarterly series were collected and processed for a Bayesian VAR model.

Time Range

1990 to 2024

Freshness

Last updated 2026-07-16 17:41:30

Geography

United States