Supplementary Material for Manuscript:Bayesian Forecasting of Financial Crises with Continuous Latent States

SUPPLEMENTARY MATERIAL FOR:"Bayesian Forecasting of Financial Crises with Continuous Latent States and Regime Ambiguity" This supplementary material accompanies the manuscript submitted to the Journal of Applied Econometrics. It contains complete replication code, supplementary figures, tables, and data descriptions necessary to reproduce all results in the paper. CONTENTS:1. Replication Instructions - Detailed setup and execution guidelines2. Supplementary Figures S1-S3 - Particle filter diagnostics, MCMC convergence checks, and Eurozone fragility analysis3. Supplementary Tables S1-S2 - Full parameter estimation results and computational benchmarks4. Complete Python Code - Self-contained replication scripts for all main text and appendix results5. Data Documentation - Sources, transformations, and simulated data generation procedures TECHNICAL DETAILS:- Language: Python 3.9+- Required packages: numpy, pandas, matplotlib, scipy, seaborn, pymc- Expected runtime: 45-60 minutes on 8-core workstation- Output formats: PDF figures, CSV tables, PNG visualizations METHODS REPRODUCED:- Bayesian estimation of continuous latent state model- Particle filter implementation for financial fragility indices- MCMC parameter estimation with convergence diagnostics- Early-warning system calibration and threshold optimization- Cross-country Eurozone fragility analysis DATA:Empirical data from public sources (BEA, BLS, FRED, Fed, CBOE, KC Fed)Simulated data generated from calibrated DSGE model parameters LICENSE:Provided for academic replication purposes. Contact the Journal of Applied Econometrics editorial office for permissions. CITATION:Please cite the accompanying paper when using this material. VERSION: January 2026JOURNAL: Journal of Applied Econometrics