Replication Data for Forecastable Fragility: 44 Markets and U.S. Policy Shocks (2010-2026)

Replication data and code for the paper 'Forecastable Fragility: Which Markets Amplify U.S. Monetary-Policy Shocks, and Why'. The dataset includes daily return and illiquidity series for up to 46 country equity markets and U.S.-dollar emerging-market sovereign bonds from January 2010 to June 2026, matched to 219 scheduled FOMC announcements. The package contains Python scripts for constructing a Transmission-Fragility Index, performing local projection estimations, and conducting cross-validation and Bayesian hierarchical estimation.

Use Cases

• Forecasting market amplification of U.S. monetary policy shocks based on the Transmission-Fragility Index.
• Analyzing cumulative return and illiquidity responses to policy announcements across 44 markets.
• Conducting leave-one-country-out cross-validation for model performance evaluation.
• Performing Bayesian hierarchical estimation of market fragility parameters.
• Replicating the paper's out-of-sample forecasting results using the provided scripts.

Strengths

• Includes data for up to 46 country equity markets and emerging-market sovereign bonds.
• Covers a time range from January 2010 to June 2026, matching 219 of 220 scheduled FOMC announcements.
• Provides a complete replication package with scripts for data construction, estimation, and inference.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect temporal bias inherent to the 2010-2026 period.

Provenance

Source

Harvard Dataverse, author Rajat Kanti Paul.

Collection Method

Data constructed from Yahoo Finance via yfinance and monetary-policy surprise series from Acosta et al. (2025).

Time Range

January 2010 to June 2026.

Freshness

Last updated 2026-07-10 14:16:25; freshness should be verified.

Geography

Up to 46 country equity markets and U.S.-dollar emerging-market sovereign bonds.