Replication Data for: Macrodynamic Interrelationships and Intersectoral Shock Transmission Effects in the Economy of Ukraine

This dataset contains all materials necessary to replicate the analysis reported in the study on shock transmission dynamics in the Ukrainian economy (2015–2023). It is the second in a two-part series of open datasets accompanying research on Ukrainian sectoral market interdependencies. This dataset is a direct continuation of: Telnova, H., Ozhelevskaya, T. (2026). Sectoral multipliers and system linkages in Ukrainian industry markets [Data set]. Zenodo. https://doi.org/10.5281/zenodo.19348878 - which provided the static characterisation of sectoral linkages. The present deposit adds a dynamic analytical layer using Panel VAR, Impulse Response Functions, and structural break testing. Coverage: 42 NACE Rev.2 subsections × 9 years (2015–2023) = 378 sector-year observations. Ukraine, excluding temporarily occupied territories. Annual Input-Output tables at basic prices. Current prices (million UAH). The dataset includes four components: (1) Raw input data (data_raw/). Three CSV files: (a) multipliers_full_2015_2023.csv - the full panel of derived sectoral indicators carried from the companion dataset: Leontief output, income, and wage multipliers; Rasmussen backward and forward linkage indices with quadrant classification; PageRank network centrality (alpha = 0.85); Systemic Impact Index (SII). (b) IO_totals_2015_2023.csv - sector-level nominal totals for gross output, gross value added, and wages. (c) gdp_deflator_2015_2023.csv - annual chain-linked GDP deflator for Ukraine (base 2015 = 100), used to convert nominal values to real 2015 prices. (2) Processed analytical tables (data_processed/). Nine CSV files: descriptive statistics by year and overall panel (T1, T1b); Im–Pesaran–Shin panel unit root test results for all five variables at level and first difference (T2); Pearson correlation matrix of first differences (T3); full PVAR coefficient table with HC1-robust standard errors and significance levels (T4), plus the compact 5×5 coefficient matrix A (T4b); IRF point estimates and 90% bootstrap confidence bounds for six horizons across three impulse-response pairs (T5); Chow F-test structural break results (T7); sector-level shock decomposition 2021–2022 with Rasmussen quadrant transition flags (T8); summary hypothesis testing table for H1–H3 (T9). (3) Figures (figures/). Thirteen PNG files at 180 dpi: median dynamics panel (Fig1); correlation heatmap (Fig2); PVAR coefficient heatmap (Fig3); three IRF plots - shocks to SII, FL, PageRank (Fig4–6); OLS-CUSUM stability test (Fig8); sectoral shock bubble chart 2022 (Fig9); Rasmussen quadrant evolution across four benchmark years (Fig10); PageRank dynamics for top-10 and bottom-10 sectors (Fig11); SII vs PageRank scatter before/after 2022 (Fig12); four-panel summary dashboard (Fig13). (4) Replication code (code/). One R script (code_analysis.R, approx. 350 lines) that reproduces all results from raw inputs to final figures and tables. Dependencies: readr, dplyr, tidyr, ggplot2, patchwork, ggrepel, sandwich, lmtest, vars, strucchange, urca. Runtime: 2–5 minutes. No proprietary software required. Key findings reproduced by this dataset: All five endogenous variables (output multiplier, SII, BL, FL, PageRank centrality) are integrated of order I(1). PageRank centrality is the dominant shock transmission channel (r = 0.667 with ΔSII). The PVAR reveals a multiplier channel (ΔO_mult → ΔSII, +0.064**; ΔO_mult → ΔPageRank, +0.110**) and a compensatory backward linkage channel (ΔBL → ΔSII, −0.224**). IRF of PageRank shocks is self-amplifying over five years. The Chow F-test confirms a structural break in 2022 (F = 3.50, p = 0.016). Six of 42 sectors changed Rasmussen quadrant in 2022. Primary data source: State Statistics Service of Ukraine. Input-Output Tables at Basic Prices 2015–2023. URL: https://stat.gov.ua/uk/datasets/tablytsya-vytraty-vypusk. Raw XLSX files from Ukrstat are not redistributed; only derived tidy CSV files and analytical outputs are included. Programming language: R (≥ 4.2.0) Licence: Creative Commons Attribution 4.0 International (CC BY 4.0)