Raw data.

Principle

All data derived from this study is public, versioned, and accompanied by provenance hash. Anyone disagreeing with an analysis can reproduce the computation from the same raw input. There's no hidden table, proprietary formula, or obscure aggregate: the study is an open notebook of computations.

License is dual: CC-BY 4.0 for data and derived text, MIT for the code that produced them. Both permit commercial use and redistribution with attribution.

Available files

V1 of the study ships seven derived files · three JSON optimized for web, three auditable CSV, and a manifest with checksums. All live in public/data/susep/ of the site repository, synced via scripts/sync-susep-data.sh from the backstage pipeline.

JSON (web · aggregates)

• agg_annual_uf_ramo.json · direct premium, incurred loss, L/P by state × line × year (655K rows after union runoff)
• agg_annual_ramo.json · national aggregation by line × year (consolidated against Σ_state)
• agg_annual_series.json · national time series by line (2014→2024 where available)

CSV (auditable · pt-BR locale)

• derived_earning_ratios.csv · loss ratio derived with union runoff, ; separators with , decimal (Excel pt-BR ready)
• fact_ganho_uf.csv · granular fact state × line × month × cogroup company (655K rows, R$ 473 bn totalled)
• tables/*.csv · 9 individual tables used in study figures, one per chart

Manifest

• MANIFEST.md · lists all files with SHA-256, extraction date, generating pipeline version, and individual license

How to use

To reproduce an analysis

# clone backstage repo (pipeline + raw)
git clone https://github.com/Natan-Cardeal/panorama-susep.git
cd panorama-susep
make reproduce
# generates public/data/susep/* identical to hashes in MANIFEST.md

To cite in academic paper

Current version is v1.0. Use the stable snapshot (commit 1e9bbbb) and reference via DOI when available in v1.1+. Before that, cite the commit hash directly.

To include in a dashboard or report

JSONs are optimized for direct fetch in browser (cache-busting via versioned filename). Use:

const data = await fetch("https://natancardeal.com/data/susep/agg_annual_uf_ramo.json").then(r => r.json())

CC-BY 4.0 license requires visible attribution · when possible, link to natancardeal.com/research in the source note.

Base updates

SUSEP updates SES typically in quarterly cycles. The backstage pipeline runs manually for each study release (v1.0 captured base updated to 2025-04-20). Future releases include:

• v1.1 · planned Q3 2025 · methodology revision + Zenodo DOI
• v1.2 · Q4 2025 · expansion to top-15 lines
• v2 · 2026 · local reinsurance (expanded SES universe coverage)

@misc{cardeal_panorama_susep_2026,
  author       = {Cardeal, Natan},
  title        = {Panorama SUSEP · Aggregated data v1.0 v1.0},
  year         = {2026},
  publisher    = {natancardeal.com},
  url          = {https://natancardeal.com},
  note         = {Accessed AAAA-MM-DD}
}

Cardeal, N. (2026). Panorama SUSEP · Aggregated data v1.0 v1.0. natancardeal.com. Retrieved YYYY-MM-DD, from https://natancardeal.com

CARDEAL, Natan. Panorama SUSEP · Aggregated data v1.0 v1.0. Curitiba: natancardeal.com, 2026. Available at https://natancardeal.com. Access on YYYY-MM-DD.