Replication package: Persistence or transmission? Measuring the horizon of financial stress transmission to euro-area equity markets
Description
Replication package for "Persistence or transmission? Measuring the horizon of financial stress transmission to euro-area equity markets" (Feki and Ellouz; version 1 of this dataset carried the paper's earlier title, "Frequency connectedness between global financial stress and euro-area equity markets"; same study, same data). The package contains everything needed to reproduce every number, table and figure: raw downloads as retrieved (OFR Financial Stress Index, daily Geopolitical Risk index, closing prices of the CAC All-Tradable, CAC 40, DAX, FTSE MIB, IBEX 35, AEX, WTI, Bitcoin and Ethereum), the strictly date-aligned daily panels for five euro-area markets (2,154 sessions, January 2018 to August 2026), the Euronext certification of the French series, the R and Python scripts in execution order, all estimation output, the figures, and complete run logs. Version 2 (15 September 2026) adds: (i) the full re-estimation behind the placebo analysis, with 500 draws of the complete and partial placebos for each market, one-ingredient placebos, a block bootstrap of the observed measures and the prewhitened systems (3,705 TVP-VAR frequency estimations); (ii) the ECB's Composite Indicator of Systemic Stress, euro-area and United States versions, in place of the OFR index, with observed, placebo and prewhitened systems and descriptive statistics; (iii) the scripts that build the manuscript directly from the result files. See README_REPLICATION_v2.md for folder contents, software versions and step-by-step instructions.
Files
Steps to reproduce
Software: R 4.6.1 with ConnectednessApproach 1.0.4, data.table, zoo, tseries, moments, gmm; Python 3.12 with pandas, numpy, requests. Unzip Replication_Package_v2.zip; every script takes the package folder (and the R library path) as arguments; paths and seeds are set in the scripts, so all results are exactly reproducible. Baseline paper (folders 01-05, about 4 hours): 1. python 02_scripts/01_telecharger.py <folder> (raw downloads; the files as retrieved are already in 01_donnees_brutes/) 2. python 02_scripts/02_certifier_cact.py <folder> (certification of the CAC All-Tradable against the official Euronext file) 3. Rscript 02_scripts/03_construire_bases.R <folder> <r_library> (date-aligned panels, 03_base_propre/) 4. Rscript 02_scripts/04_descriptives_gmm_gbm.R (descriptive statistics, unit-root tests, GMM) 5. Rscript 02_scripts/05_tvpvar_baseline.R (time-domain and frequency connectedness, robustness) 6. Rscript 02_scripts/06_extension_europe.R (five euro-area markets) 7. Rscript 02_scripts/08_placebo_persistance.R, 10_placebo_decomposition.R, 14_placebo_complet.R (persistence placebos) 8. Rscript 02_scripts/11_figures.R Re-estimation with 500 draws, block bootstrap, prewhitening (folder 06, about 6 hours on 11 cores): 9. bash 06_reestimation_2026-09-13/02_scripts/lancer_tout.sh 11 10. python 06_reestimation_2026-09-13/02_scripts/agreger.py 06_reestimation_2026-09-13 ECB CISS in place of the OFR FSI (folder 08, about 2.5 hours on 12 cores): 11. python 08_ciss_2026-09-14/02_scripts/build_bases.py 08_ciss_2026-09-14 <folder> 12. bash 08_ciss_2026-09-14/02_scripts/lancer_ciss.sh 12 13. python 08_ciss_2026-09-14/02_scripts/agreger_ciss.py 08_ciss_2026-09-14 14. Rscript 08_ciss_2026-09-14/02_scripts/descriptives_ciss.R 08_ciss_2026-09-14/03_base <r_library> 08_ciss_2026-09-14 Manuscript (folder 09): python build_jfs.py reads every CISS number from 08_ciss_2026-09-14/04_resultats/ and writes Article_JFS.tex; compile twice with pdflatex (elsarticle class). All estimation output and the summary tables are included, so every table can be checked without re-estimating. See README_REPLICATION.md for details.
Institutions
- Faculty of Economics and Management of SfaxSfax Governorate, Sfax