About the study.

Why this study exists

SUSEP has published statistical reports since 1998. They're dense files, in technical CSV format, with line nomenclature that changed three times in the last decade, and almost no visual analysis. Anyone who needs to look at the Brazilian insurance market in depth has three options: pay private consultancy (R$ 50K–200K per report), submit a request under FOI law and wait two months, or download the CSVs and build everything from scratch.

This notebook does the third path, in public.

The intent is not to replace consultancies · it's to open a base layer of honest primary reading that makes it harder for anyone (including the author) to invent narrative on top of numbers nobody knows the origin of.

Authorship

Natan Cardeal Rodrigues, Curitiba PR, actuary-in-training. Full-time, I work as a Data Specialist at Jones (insurtech based in New York · COI compliance for commercial real estate in the USA). In parallel, I maintain other projects that do not operate in the Brazilian insurance market studied in this notebook. No affiliation with Brazilian insurer, broker, reinsurer, or regulator. The study was written during 2025 in hobby-time, self-funded.

For complete biographical details, see /sobre.

How to cite

V1 keeps short form without DOI. Use:

@misc{cardeal_panorama_susep_v1,
  author = {Cardeal, Natan},
  title  = {Panorama SUSEP · v1.0},
  year   = {2025},
  url    = {https://natancardeal.com/research},
  note   = {Accessed on [date]}
}

In running text (ABNT NBR 6023):

CARDEAL, Natan. Panorama SUSEP · v1.0. Curitiba, 2025. Available at: natancardeal.com/research. Accessed on: [date].

Each interactive slice (specific map) has its own URL via query string, which can be cited directly. The "§ cite this slice" block in /research/panorama-uf/ generates the citation with the current selection URL in BibTeX, APA, and ABNT.

Editorial convictions

Data is not argument. It's raw material. The argument is what you do with it: which slices you choose, which you refuse, which caveats you state out loud. The honesty of an analysis is measured by the size of its caveat, not by the strength of its conclusion.

Analysis without an auditable method is opinion in a nice suit. Everything I publish here carries a provenance sheet: primary source, extraction date, the exact script that produced the aggregate, hash of the raw file. Anyone who disagrees can reproduce, flag the error, and I correct it in versioned errata. It's slow. It's the opposite of viral.

Descriptive statistics is routinely underestimated. Before any model, before any forecast, there's the work of looking at the data and understanding its shape: where it concentrates, where it's missing, where it lies about itself. Much of what passes for sophistication is anxiety about skipping that step.

Publish early, publish with caveats, publish with open errata; it beats publishing late, definitive, and being wrong with no correction mechanism. The notebook is, before anything else, a public contract that I can be contested.

Brazil's insurance sector is poorly explained by the press and understudied in academia. There's a wide window between what the regulator publishes (massive CSVs, technical jargon) and what reaches the general reader. Closing that window, with method, with provenance, and without flattening nuance, is the project.

Editorial commitment

This notebook operates under five non-negotiable principles:

Voice and style

The study's voice inherits the tradition of serious descriptive statistics from before Microsoft Excel · third-person impersonal in analytical body, implicit second person in interactive controls, technical vocabulary explained at first use. No self-promotion, no superlatives ("revolutionary", "robust", "unique"), no emoji, no exclamation.

The complete voice-guide (8 pages) is available in editorial format in the backstage repository, at docs/voice-guide.md (pt-BR) and docs/voice-guide-en.md (en).

Detailed licensing

Component	License	Permitted use
Editorial text (this site)	CC-BY 4.0	Commercial, derivative, redistribution with attribution
Derived data (public/data/susep/)	CC-BY 4.0	Same
Pipeline code (backstage repo)	MIT	Commercial, derivative, sub-licensing
Raw SUSEP data	Public domain (Brazilian FOI law)	No restriction

The MIT + CC-BY 4.0 combination covers 99% of use cases (academic work, institutional dashboard, data journalism). For cases where one of the licenses doesn't fit (e.g., work that needs relicensing incompatible with CC-BY), write to me.

§ procedência

fonte

natancardeal.com · about the study

extração

2025-04-24

pipeline

editorial notebook · v1.0

licença

CC-BY 4.0

backstage repository ↗