raisr: RAIS microdata from R • raisr

raisr downloads and reads the public, non-identified microdata of the RAIS (Relação Anual de Informações Sociais), the annual census of formal employment relationships and establishments published by the Brazilian Ministry of Labour and Employment on the PDET FTP server.

Function	What it does
`rais_available()`	Lists every archive published on the server, with size, date and edition (final, partial, legacy)
`rais_files()`	Resolves, offline, which regional or state archive holds a given state in a given year
`rais_download()`	Downloads those archives into an idempotent local cache
`rais_read()`	Reads a `.7z` archive as a stream, filtering by state and selecting columns before anything is kept in memory
`rais_fetch()`	Download and read in one call, several years at once
`rais_stock()`	Employment stock on 31 December, admissions, separations and December payroll by any grouping
`rais_layout()`	The official record layout: every column, its two header spellings, type and meaning

The files are large: from the RAIS 2018 onwards, the employment relationships come in one regional archive per group of states (about 130 MB compressed for the North, 600 MB for the Northeast and more than 1 GB for Sao Paulo, each with tens of millions of records). Reading one state out of a regional archive takes a few minutes and well under 1 GB of memory, because the archive is decompressed on the fly and filtered chunk by chunk.

❕ Disclaimer
This package reads public data published by the Brazilian Ministry of Labour and Employment (MTE) through the PDET programme, which is the institution responsible for the data and its documentation. The files, their layout and their publication schedule are the Ministry’s; this package only automates listing, downloading and reading them. Column names are the Ministry’s Portuguese headers normalized to snake_case without accents (municipio, vinculo_ativo_31_12, vl_remun_media_nom); see rais_layout() for their meaning. The official layouts are published in the same FTP folder as the data: ftp://ftp.mtps.gov.br/pdet/microdados/RAIS/.

Installation

# From CRAN (when available):
install.packages("raisr")

# Development version:
# remotes::install_github("StrategicProjects/raisr")

raisr reads .7z archives through the archive package (libarchive). Binaries for Windows and macOS bundle it; on Linux install libarchive-dev (Debian/Ubuntu) or libarchive-devel (Fedora) first.

Quick start

library(raisr)

# What is on the server for 2024?
rais_available(year = 2024)

# Which archive holds Pernambuco in 2024? (offline)
rais_files(2024, uf = "PE")

# Pernambuco (IBGE code 26), RAIS 2024, a few columns.
# Downloads the NORDESTE regional file (about 600 MB) once, into the cache.
Sys.setenv(RAISR_CACHE_DIR = "~/dados/rais")
pe <- rais_fetch(2024, uf = "PE",
                 columns = c("municipio", "cnae_20_subclasse", "vinculo_ativo_31_12",
                             "mes_admissao", "mes_desligamento", "vl_remun_dezembro_nom"))

# Stock on 31/12, admissions, separations and December payroll by municipality
rais_stock(pe, by = "municipio")

# Several years stack on the same column names, whatever the header generation
pe_serie <- rais_fetch(2019:2024, uf = "PE", columns = c("municipio", "vinculo_ativo_31_12"))
rais_stock(pe_serie)

# Establishments instead of employment relationships
est <- rais_fetch(2024, uf = "PE", type = "estabelecimentos")

Everything that reads data can be tried offline with the sample archives shipped in inst/extdata:

f <- system.file("extdata", "2024", "RAIS_VINC_PUB_NORDESTE_sample.7z", package = "raisr")
rais_read(f, uf = "PE")

How the server is organized

ftp://ftp.mtps.gov.br/pdet/microdados/RAIS/
    1985/ ... 2017/      one file per state (PE2017.7z, SP2017.7z, ...) + ESTB2017.7z
    2018/ ... 2025/      RAIS_VINC_PUB_{NORTE,NORDESTE,MG_ES_RJ,SP,SUL,CENTRO_OESTE,NI}.7z
                         + RAIS_ESTAB_PUB.7z
    2023 Parcial/        preliminary edition of a year, when one was published
    2023/Legado/         the previous files, kept when a year is re-published
    Layouts/             the official layout spreadsheets

rais_files() encodes this routing, rais_download() mirrors it in the cache (<cache>/2024/RAIS_VINC_PUB_NORDESTE.7z, <cache>/2023-legado/...) and rais_read() takes the reference year from the folder or the file name.

Two header generations

The Ministry changed the file format with the RAIS 2023 (re-published in 2026) and the years after it: from ;-separated text with decimal comma and headers such as Vínculo Ativo 31/12 to ,-separated text with decimal point and headers such as Ind Vínculo Ativo 31/12 - Código. The partial and legacy editions still use the older format. rais_read() detects the format from the header and maps both generations to the same normalized names, so rais_fetch(2019:2024, ...) stacks without further work. rais_layout() lists the two spellings of every column.

Network requirements

The server is an FTP server. Outbound access to ftp.mtps.gov.br on port 21 and to the high ports used by passive mode is required; corporate firewalls often block one or both. rais_available() returns an empty tibble with a warning when the server cannot be reached.

cagedr does the same for the monthly Novo CAGED files of the same server, with the same conventions.
basedosdados gives SQL access to a treated copy of the RAIS in BigQuery; raisr reads the Ministry’s files directly and needs no account.
tesouror, comexr and pixr cover other Brazilian public data sources with the same conventions.

License

MIT, see LICENSE.md.

raisr