Transition Risk Index for EU Regional Manufacturing

A composite indicator measuring how exposed and vulnerable EU NUTS-2 regions are to the costs of industrial decarbonisation. Covers 237 regions across the EU-27, disaggregated by 10 manufacturing subsectors.

The TRI follows the IPCC AR6 risk framework: TRI = Exposure^0.5 x Vulnerability^0.5

Repository structure

Regio-Decarb-Systemic-Framework/
|
|-- README.md
|-- .gitignore
|
+-- Code and data/
    |-- _targets.R               # Pipeline definition (targets)
    |-- R/                       # Function files (sourced by targets)
    |-- Code/                    # Standalone scripts (legacy/reference)
    |-- Initial data/            # Input datasets
    |-- Derived data/            # Intermediate outputs (reproducible)
    |-- Final data/              # Risk_data.xlsx, Sensitivity_Analysis.xlsx
    +-- Figures/                 # Output figures (PNG, 600 DPI)

How to reproduce

Requirements: R >= 4.5, internet connection for Eurostat API calls on first run.

External data (not auto-downloaded)

Four files must be downloaded manually and placed under Code and data/Initial data/:

File	Source
Non sector data/qog_eureg.csv	QoG EU Regional Dataset
Non sector data/qog_ei_eureg.csv	Same website
Non sector data/ENSPRESO_Integrated_Data/ENSPRESO_Integrated_NUTS2_Data.csv	JRC ENSPRESO
Non sector data/TECH-RIS.xlsx	EC Regional Innovation Scoreboard — annual bulk Excel, no stable URL. Place the latest release at this path.

All other inputs are auto-downloaded from Eurostat at pipeline run-time.

How years are chosen

Every indicator independently selects the latest year for which Eurostat has complete EU-27 coverage (the threshold is per-country and, for sector indicators, per (country × NACE sector)). The chosen years can differ across indicators — e.g. Unemployment may be 2024 while FIGARO-derived Scope 3 may be 2023. After each tar_make() run, Final data/Coverage_Report.xlsx lists the year each indicator picked along with its Eurostat source dataset.

If Eurostat publishes new data, simply re-run tar_make() (after tar_destroy() if you need to invalidate the cache); the year selector will pick up the new vintage automatically.

Downscaling methodology (consistent across all national-source indicators)

• Extensive quantities (€, kt CO2eq, GWh, persons, count) → employment-share downscaling. For each (Country, Sector) the national value is split across the country's NUTS-2 regions in proportion to that region's share of national manufacturing employment in that sector.
• Intensive quantities (ratios, indices, percentages) → uniform replication. Each NUTS-2 region of a country receives the same national value.

Run the pipeline

setwd("Code and data")
library(targets)
tar_make()

Reproducibility model

The committed xlsx files in Code and data/Initial data/ and Code and data/Final data/, together with the PNGs in Code and data/Figures/, are the published snapshot. Anyone who clones the repo at the paper's submission commit (or tag) sees the exact numbers that appear in the manuscript without re-running anything.

Re-running tar_make() overwrites those files with whatever vintage Eurostat currently publishes. The pipeline is intentionally rolling: every indicator picks its own latest year with ≥95% EU-27 NUTS-2 coverage (see Coverage_Report.xlsx after each run). The git diff between two commits is then the audit trail of how the rolling vintage moves over time.

For a strict replication of the paper:

1. git checkout <submission-tag> (or the relevant commit hash)
2. Read the xlsx files directly. No tar_make() needed.

For an updated run against the latest Eurostat data:

1. tar_destroy() to clear the cache
2. tar_make() to refetch and rebuild everything
3. Inspect Final data/Coverage_Report.xlsx to see which year each indicator picked.