penality-analysis

Paper: Cycling poverty in France: a triple-penalty mobility-justice diagnostic on 34,858 communes. Rohan Fossé and Gaël Pallares, CESI LINEACT, 2026.

This repository packages a mobility-justice diagnostic that turns the IMD-4 cycling-environment indicator into an actionable priority list for the second tranche of the French Plan Vélo (2023–2027). It is one of five repositories in the cycling-data-lab GitHub organisation.

Naming note. The repository is named penality-analysis (with the typo) to keep stable URLs once cited; the paper itself uses the correct triple-penalty terminology throughout.

TL;DR

The standard French cycling-policy diagnostic answers where the cycling environment is weak. It does not answer the policy question that follows — who pays the cost of that weakness. We close this gap by stacking three vulnerability layers on top of the commune-level IMD-4:

1. Cycling-environment deprivation (bottom-33% IMD-4)
2. Monetary-poverty exposure (top-33% INSEE poverty rate, > 15%)
3. Structural geographic isolation (lowest income tertile ∪ overseas départements)

The triple-penalty intersection (the product of the three indicator flags) identifies:

• 322 of the 362 cycling-poverty deserts as also monetary- poverty-vulnerable, covering 1.89 M inhabitants (91.7% of desert population).
• 42 of the 362 deserts in the overseas départements — 11.6% of the count but 36.3% of the population — making Outre-mer a structurally over-represented mobility-justice priority.
• 90 metropolitan deserts in the national first income decile but outside Outre-mer, aggregating to ~450,000 inhabitants and invisible under any single-axis equity filter. This is the policy-blind subset of the current Plan Vélo distribution.

Build the paper

pdflatex imd_social.tex
bibtex   imd_social
pdflatex imd_social.tex
pdflatex imd_social.tex

Compiles cleanly on TeX Live 2026.

Reproduce the analysis

The triple-penalty diagnostic is a transparent overlay of three public data layers:

• IMD-4 : from the parent repository imd-national-catalogue.
• Income median, poverty rate : INSEE Filosofi commune-level open data, most recent vintage.
• Outre-mer status : binary flag from INSEE commune code (department code starts with 97).

No model fitting is required — the three flags are deterministic thresholds (Section "Method" of the paper). The computation is sub-second on a modern laptop and is documented inline.

Companion papers

• imd-national-catalogue — the IMD-4 indicator (substrate of this analysis).
• bikeshare-demand-forecasting — predictive paper on the IMD-4.
• bikeshare-gsp-tools — graph-signal-processing companion paper.
• gbfs-audit-catalogue — the underlying station inventory.

How to cite

@unpublished{FossePallares2026cyclingPoverty,
  author = {Foss\'e, Rohan and Pallares, Ga\"el},
  title  = {Cycling Poverty in {F}rance: A Triple-Penalty
            Mobility-Justice Diagnostic on 34{,}858 Communes},
  note   = {CESI LINEACT, 2026.
            \url{https://github.com/cycling-data-lab/penality-analysis}},
  year   = {2026}
}

A machine-readable CITATION.cff is provided.

License

MIT. The underlying INSEE Filosofi data is published under the Licence Ouverte 2.0 (Etalab).

Contact

Rohan Fossé — rfosse@cesi.fr — ORCID Gaël Pallares — ORCID