Six years of formally registered employment in Brazil, from the pandemic collapse to the 2025 slowdown, set against a long-run trend reaching back to 2007, with forecasts to mid-2027 and a data dossier on the proposed end of the 6×1 work schedule. Built end to end from official CAGED microdata.
An editorial, data-driven report on the Brazilian formal labor market, the CLT universe of fully-registered jobs, drawn directly from the official CAGED microdata. It reads the last six years as one continuous story: the pandemic collapse, the V-shaped rebound, and the cooling expansion of 2024–2026, then projects the national and per-state series through June 2027. A long-run trend line extends the picture back to 2007, using the legacy CAGED to frame the recent cycle against two decades of formal employment.
The closing chapter is a dossier on the most consequential labor debate in Brazil since 1988: the proposal to replace the 6×1 schedule (six workdays, one rest day) with 5×2. Using contracted-hours data, it maps which sectors are most exposed and weighs the proposal against peer-reviewed economic literature.
The report is fully interactive and self-contained, with sortable tables, a per-state explorer and confidence bands, published in Portuguese, English and Spanish. This page is the summary; the buttons above open the live version.
The net balance reproduces the official "unadjusted" Novo CAGED series exactly, as the sum of net movements per reference period. The core analysis and all forecasts use the Novo CAGED (eSocial-based) from January 2020 onward; the pre-2020 trend line is built from the legacy CAGED (Law 4.923/65 declarations, 2007–2019), shown only as long-run context. The January 2020 switch from the old declarations to eSocial is a methodological break, so the two eras are not strictly comparable and the report marks that boundary explicitly. A handful of legacy months whose MTE archives are corrupt are filled by time interpolation and flagged in the report. Working time is read from the horascontratuais field, with the 44-hour week used as a proxy for the 6×1 universe. Forecasts come from classical and machine-learning time-series models, selected per series by RMSE in a 12-month holdout, with explicit diagnostics for seasonality and heteroskedasticity.
The literature review behind the dossier was assembled through a scientific paper-lookup protocol over the OpenAlex base, with no fabricated references: every cited study carries a verifiable DOI. The full code and data pipeline are open source.
The interactive edition opens in a new tab. Pick a language: