Open IBGE microdata (PNAD Contínua) with the SAS input layout
IBGE publishes PNAD Contínua, Census and POF microdata as one large fixed-width TXT: no delimiter, no header, just digits. Where each variable sits is described in a separate SAS input script (and, for several surveys, an SPSS one). If you work in Python, Excel or Power BI rather than R or Stata, that script is the missing piece — and it is all you need to turn the TXT into a table.
Free to use. The file is read in chunks inside your browser tab and never uploaded. Windows version — 4.9 MB, no installation: details. Pro — $29 once: price.
What to download
- Data: the quarter’s ZIP with one TXT inside (for example
PNADC_012023.txt, a few hundred megabytes). - Documentation: it contains the SAS input — a file such as
input_PNADC_trimestral.txt(sometimes.sas) with lines like@0001 Ano $4. /* Ano de referência */— and the variable dictionary in Excel.
Open it
- Select both files at once — the data TXT and the SAS input (or the SPSS
.sps) — and drop them on the window. - The input is recognised by its content even with a
.txtextension, and the data is split by it: every variable at its position, named as in the script (Ano,UF,V2007,V2009,V1028…). - Already opened the data alone? In “Parsing”, click “Use a layout from a SAS/SPSS file…”, or drop the input onto the open tab.
Variables declared with $ stay text, so state and municipality codes keep their leading zeros; the rest become numbers, implied decimals such as 8.2 are applied, and a lone . is a missing value. The Portuguese comments next to each variable (or LABEL statements) become column labels, so “Find column…” (Ctrl+G) finds V2009 when you type “idade”.
What you can do with it
- Filter a state, an age band or an employment status and export just that subset to CSV, Excel or Parquet.
- Sum the weight
V1028byUFin a pivot table to get estimated population per state. Design-based variance and confidence intervals remain a job for R’s survey package or Stata.
The TXT is indexed in one pass and rows are read on demand, as with any large CSV. Category labels (1 = Homem, 2 = Mulher) are not applied — the SAS input does not contain them; keep the dictionary at hand. The free edition exports up to 5,000 rows per file; Pro ($29, one-time) removes the limit.
FAQ
Does it work for the Census, POF or the old PNAD?
Yes, whenever the survey ships a SAS input in the @position name $width. form or an SPSS DATA LIST with column ranges. Only that part of the script is read; nothing is executed.
Is the data uploaded anywhere?
No. The file is read in your browser or in the desktop app on your computer.
Can I edit the TXT and save it in the same layout?
No — with a SAS/SPSS layout the table is read-only, because bytes between variables are not read. Save the result as CSV, Excel or Parquet.
Free to use. Your file is not uploaded to a server. Windows version — 4.9 MB, no installation: details. Pro — $29 once: price.