Macro indicators: dollar and euro (PTAX), Selic, IPCA, GDP and unemployment

Six official indicators, from two public government sources, kept as the source publishes them, with no recalculation: the PTAX dollar and euro rates and the Selic rate, from the Central Bank of Brazil; IPCA inflation, GDP and the unemployment rate, from IBGE. Each one has a page under /macro/..., a JSON/CSV route under /macro/series/... and a ready-made citation line.

Why they live in the Macro section

Every company figure in this database is recomputable: download the CVM filing and you reach the same value. These are not. They are aggregates compiled by the Central Bank and by IBGE that nobody rebuilds from a public document — you either accept the official figure or you don't use it. So they live in their own table (macro_serie) and their own section, with the source and the licence written on every page and in every response. The same rule as the foreign flow.

The series, and how the codes were chosen

The codes were checked against each source's official documentation, by name, never by trial and error: getting a code wrong means publishing another series while believing it is this one.

Central Bank — SGS (api.bcb.gov.br), names checked in the dadosabertos.bcb.gov.br catalogue:

Series 1 and 10813 are the PTAX closing rate. The euro ones (21619 and 21620) are not listed in the catalogue as a dataset of their own; they were checked against the Closing bulletin of the Central Bank PTAX API (the "Exchange rates - all daily bulletins" dataset): on 2026-10-27, 28 and 29 the values are identical (2026-10-29: ask 6.0000, bid 5.9988).

IBGE — SIDRA (apisidra.ibge.gov.br):

For GDP, both tables are filtered by classification 11255 = 90707 (GDP at market prices). Changes are in volume (net of inflation); the value in reais is at current prices.

The licence of each source

Checked before publishing, as we did with the CVM.

• Central Bank: every dataset on the BCB open data portal declares the ODbL 1.0 licence (Open Data Commons Open Database License) — the same as the CVM. Free use, including commercial, citing the Central Bank as the source and sharing the derived database on the same terms.
• IBGE: IBGE statistics are public and free to reproduce with the source cited ("IBGE"); they are federal open data (Decree 8,777/2016). We found no restriction on redistribution or on commercial use.

Left out by decision: the CDI (computed by B3; we use the Selic as the reference), the IGP-M (FGV) and the ANBIMA indices — all with usage rules of their own that we have not checked.

Since when

• Dollar, euro and Selic: since 1999 (the first year of the floating exchange rate and of the Copom Selic target).
• IPCA: since July 1995, the first month whose 12-month window fits entirely inside the Real Plan. Before that the cumulative figure carries the 1994 hyperinflation.
• GDP: since the first quarter of 1996, the start of tables 5932 and 1846 on SIDRA.
• Unemployment: since the rolling quarter ending in March 2012, the start of PNAD Contínua.

How collection works

• Every week, when the database is rebuilt, each series is downloaded in full. The SGS API refuses daily queries spanning more than 10 years, so the period is split into 9-year windows, with a retry when a window fails.
• Every night, together with the day's prices, only the recent part: the last 45 days of the daily series and the last 24 months or quarters of the IBGE ones. This keeps PTAX and Selic current between Sundays.
• Isolated failure. Each series is downloaded, checked and written on its own. If the Central Bank or IBGE is down, that series stays as it was and the rest of the database goes on. Nothing is deleted; a from-scratch rebuild inherits from the previous database whatever the source did not return.
• Plausible range. Each series has a wide range (PTAX between R$ 0.50 and R$ 20, Selic between 0 and 100% a year, monthly IPCA between -5% and 10%...). A value outside it is not economics, it is a misread — swapped decimal separator, unit, shifted column — and the whole batch is rejected, with nothing written.
• On /saude, the macro block shows the date of the latest figure of each series, how many days have passed since the end of that period and the limit for each frequency (7 days for daily series, 50 for IPCA, 70 for unemployment, 170 for GDP), plus the result of the latest collection.

Checks

• Ask never below bid in the dollar and euro PTAX.
• 12-month IPCA equal to the product of the 12 published months, within 0.02 percentage point (IBGE rounds each month to two decimals).
• IBGE IPCA equal to the Central Bank one: the SIDRA month by month matches SGS series 433, which the database already used for the real return.

What the page does NOT do

• It does not recompute or smooth anything: the number is the published one. The only arithmetic is the change between two figures of the source itself (the dollar on the day and over 12 months), and the page states the base.
• It does not track revisions step by step: IBGE revises GDP in later releases, and the database keeps the most recent version of each quarter.
• It is not investment advice.

Access

• Pages: /macro/dolar, /macro/euro, /macro/selic, /macro/ipca, /macro/pib, /macro/desemprego. Open.
• /macro/series — the catalogue in JSON, with the coverage, the source and the licence of each series.
• /macro/series/{codigo} — the series in JSON or CSV (?formato=csv), cut by de and ate. The daily ones (dolar, euro, selic) return, without a key, a sample of the last 30 days, and the full series comes with the free key — the same rule as the index series. The monthly and quarterly ones (ipca, pib, desemprego) come in full, open. Every response carries the fonte, the licence and the citacao.