Brazil Discount Rate: Monthly IMF Data Since 1996, and How It Differs from the Selic

This dataset is the Brazilian discount rate as published by the International Monetary Fund in its International Financial Statistics and redistributed through FRED as series INTDSRBRM193N. It runs monthly from October 1996 across 358 observations, from an all-time high of 49.75% in October 1998 to a low of 8.014% held through the winter of 2020 and 2021. It is not the Selic Over, the daily overnight rate that the Banco Central do Brasil publishes as its policy benchmark, and the section on common pitfalls below sets out what is known and what is not known about the difference.

Dataset: Brazil Discount Rate (1996–2026) · Updated 2026-08-01

Source: International Monetary Fund, International Financial Statistics · via FRED series INTDSRBRM193N

Macro Takeaway

Dataset Overview

Dataset Variables

The CSV, Excel and JSON files contain the following columns.

Column names match the CSV headers exactly. The column is named selic_rate for historical reasons in the Eco3min pipeline; the value it carries is the IMF discount-rate series described above, not the Banco Central do Brasil Selic Over.

Download the Complete Dataset

The full Brazil discount rate dataset is available in CSV and Excel formats.

FRED Direct CSV Access

The underlying data is available from FRED under series code INTDSRBRM193N:

https://fred.stlouisfed.org/graph/fredgraph.csv?id=INTDSRBRM193N

Direct CSV Access – Eco3min Structured Dataset

https://eco3min.fr/dataset/brazil-selic-rate.csv

This URL returns the complete dataset in CSV format. It can be used directly in pandas, R, curl, or any data tool.

Using the Dataset in Python

import pandas as pd

url = "https://eco3min.fr/dataset/brazil-selic-rate.csv"
df = pd.read_csv(url, parse_dates=["date"])

print(df.tail())
print(df["selic_rate"].describe())

Using the Dataset in R

library(readr)

url <- "https://eco3min.fr/dataset/brazil-selic-rate.csv"
df <- read_csv(url)

tail(df)
summary(df$selic_rate)

Both examples load the dataset directly from the URL, with no download or API key required.

Methodology

The series is compiled by the International Monetary Fund for its International Financial Statistics database, where it appears under the discount-rate heading for Brazil, expressed as a percent per annum and not seasonally adjusted. The Eco3min pipeline pulls it from the FRED API and writes the CSV, Excel and JSON files, rebuilding the full history on each run rather than appending, so any upstream revision propagates automatically.

What the IMF does not publish, in this case, is the method. FRED’s own notes for INTDSRBRM193N state that the documentation sits in the IMF’s International Financial Statistics yearbooks, that the Saint Louis Fed has requested those publications, and that notes will populate once they become available. As of September 2026 they have not. That absence is the reason this page describes the series by its published label and by its observable behaviour rather than by a definition it cannot source.

Data Quality & Provider Notes

The series is undocumented at source. Neither the IMF entry nor the FRED page carries a methodological note explaining which Brazilian instrument the discount rate refers to, how it is sampled within the month, or whether the convention has changed over thirty years. Anything asserted beyond the published label would be inference, and this page does not make it.

It behaves like an administered rate. The internal evidence is unambiguous even without documentation: the value 20.999 appears in sixteen months, including fourteen consecutively from August 2015; 8.014 appears in six consecutive months from September 2020; 21.794 in eight consecutive months from July 2025. A market-determined rate does not print identical values for over a year. The series records decisions held between changes.

Licence. FRED marks this series “Copyrighted: Citation Required”, with the copyright held by the International Monetary Fund. The IMF terms permit redistribution, including of data obtained from another party, on three conditions: attribution to the IMF with a link to the database, preservation of the integrity of the data with any material transformation declared, and transmission of the same conditions to downstream users. That last condition is why this dataset is not offered under a Creative Commons licence: a CC BY grant would tell the downloader the opposite of what the IMF requires. Eco3min applies no transformation to the values.

Alternative sources. The Banco Central do Brasil publishes its own official rate series, including the Selic, through the SGS time-series system. That is the authoritative source for Brazilian policy rates, and the one to use when the Selic itself is what is needed.

Common Pitfalls When Using This Series

1. It is not the Selic Over. The Selic Over is the daily average overnight rate on operations collateralised by federal securities, published by the Banco Central do Brasil, and the Copom sets a target for it at each meeting. This series is a monthly IMF discount-rate figure. The two are related in the sense that both describe the price of short-term money in Brazil, but they are different instruments measured at different frequencies, and their levels differ. A page or a model that needs the Selic should take it from the Banco Central do Brasil, not from this file.
2. Reading the nominal level as tightness. A 20% nominal rate in Brazil and a 5% nominal rate in the United States do not describe comparable monetary stances. Brazilian inflation history, the term premium attached to it, and the currency risk embedded in local rates all sit inside that number. Comparisons across countries require deflating by each country’s own inflation, and even then the exercise is imperfect.
3. Treating the late-1990s peaks as a policy stance. The readings above 40% between November 1997 and March 1999 were a currency defence under a crawling peg, not a domestic inflation-targeting decision. Brazil adopted inflation targeting in 1999, after the float. Averaging across that break produces a mean that describes neither regime.
4. Assuming monthly granularity reflects decision dates. Copom meets roughly every six weeks, and this series reports monthly. A month containing a decision blends the periods either side of it, which is one reason the monthly plateaus do not align exactly with announced changes.

Historical Regimes

1996–1999 – Defending the real. The series opens at 25.78% in October 1996 and averages 31.00% through 1999. Nine months print above 40%, all between November 1997 and March 1999: 46.44% in November 1997 as the Asian crisis spread, 42.34% in September 1998 after the Russian default, the all-time high of 49.75% in October 1998, and 46.00% in March 1999 following the January float. The rate is functioning as a currency defence.

1999–2002 – The unwind, and the fastest declines in the record. Once the peg was abandoned the rate fell faster than it has at any other point: down 28.46 points in the twelve months to October 1999 and 24.78 points in the twelve months to March 2000. The 2002 election cycle interrupted the descent, taking the rate back to 30.42% in December of that year.

2003–2013 – Inflation targeting settles in. The 2000s average 22.03% and the 2010s 16.48%, a structural step down as the targeting framework accumulated credibility. Waypoints: 26.78% in May 2005, 15.55% in July 2009 after the global financial crisis, and 13.58% in March 2013, at the time the lowest reading in the series.

2015–2016 – Recession and the long plateau. Through a deep domestic recession the rate was raised and then held at 20.999% for fourteen consecutive months, from August 2015 to September 2016, the longest unchanged stretch in the record.

2017–2019 – Disinflation and easing. With inflation falling the rate came down to 12.97% by March 2018, the low of the pre-pandemic era.

2020–2021 – The pandemic floor. Fifteen months print below 10%, from April 2020 to June 2021, and the series reaches its all-time low of 8.014%, held from September 2020 through February 2021. Nothing in the previous twenty-four years comes close.

2021–2026 – Back up, and holding high. From 8.01% in February 2021 the rate reached 15.29% by December of the same year, a 7.3 point rise in ten months, then 20.39% by August 2022. It eased to 17.02% by June 2024 before returning to a plateau of 21.794% held from July 2025 to February 2026, and stands at 20.999% in July 2026, the 63rd percentile of its own history.

Related Macroeconomic Datasets

An emerging-market short rate reads against its currency, against the dollar anchor, and against US inflation.

Macroeconomic Dataset Hub

This dataset is part of the Eco3min macro-financial data repository.

Explore the Eco3min Dataset Hub

Sources

• International Monetary Fund – International Financial Statistics, discount rate for Brazil. Reprinted with permission. Terms of use at https://www.imf.org/external/terms.htm
• Federal Reserve Bank of St. Louis – FRED series INTDSRBRM193N, redistributor of the IMF series
• Banco Central do Brasil – SGS time-series system, the authoritative source for the Selic policy rate
• Eco3min Research – structured dataset compilation

Dataset Reference

Eco3min Research. (2026). Brazil Discount Rate, Monthly IMF Data (1996–2026).<br> https://eco3min.fr/en/brazil-selic-rate-dataset/