The minimum wage buys three times as much as in 1995. Most of the gain came before 2012

In January 1995, six months after the arrival of the real, the minimum wage was worth the equivalent of R$ 528 today. In September 2026, it is worth R$ 1,621. After inflation, it buys 3.07 times as much.

The gain came at two speeds. From January 1995 to March 2012, the minimum wage's purchasing power multiplied by 2.62. From March 2012 to now, it rose 17%.

Set beside the rest of the economy, the number takes shape. The minimum wage rose more than GDP per capita, kept pace with the average wage over the past decade and, in dollars, is worth less than it was in 2012. The data come from IBGE, the Central Bank and Ipea (the government's applied economic research institute). At the end of the text, a series explorer and a calculator let you redo the numbers.

How we measure purchasing power

Purchasing power here means the minimum wage adjusted by the INPC, IBGE's price index for lower-income households. It is the same index the law uses to restore inflation in the annual adjustment. All values are in September 2026 reais. Saying that the January 1995 minimum wage is worth R$ 528 means that wage would buy today what R$ 528 buys.

People who live on the minimum wage feel the raise once a year and inflation the rest of the time. Between one adjustment and the next, the amount on the paycheck stays fixed while prices rise, and the money buys a little less each month. That is why the chart line has a sawtooth shape: it jumps at the adjustment and slowly slides down until the next one.

The size of each tooth depends on inflation. From 1996 to 2025, the INPC rose most in 2002 (14.7%) and least in 2017 (2.1%). In 2002, the minimum wage was worth R$ 838 in April and ended the year at R$ 750, 10.5% less in eight months, without the amount on the paycheck changing. In 2017, it went from R$ 1,487 in January to R$ 1,463 in December, a loss of 1.6%.

Looking only at the month of the adjustment shows the wage at its best. That is why, whenever possible, the comparisons that follow use every month of the year or the December-to-December change.

Year by year

Counted from December to December, the minimum wage gained purchasing power in 25 of the 31 years between 1995 and 2025. It lost in six: 1999, 2002, 2015, 2018, 2020 and 2021. The biggest jumps were in 1995 (17.1%), 2006 (13.5%) and 2005 (9.8%).

The calendar also changed. Until 2009, the adjustment came between February and May. Since 2010, it comes in January, with extra increases in March 2011, February 2020 and May 2023. The lowest point in the whole series is April 1995: R$ 502 in today's money, on the eve of that year's adjustment.

The rule in force is the one in Law No. 14,663 of 2023. The January 1 adjustment restores the INPC for the 12 months through November and adds the real growth of GDP from two years earlier. If GDP fell, only the inflation restoration remains. In January 2027, GDP for 2025 counts, which grew 2.29% by the IBGE series used here. A Diário do Comércio report from March 2026 estimated a real gain of 2.3%, within the 0.6% to 2.5% band imposed by the fiscal framework.

The minimum wage and the economy

From August 2001, the first month for which the calculation can be made with these data, to May 2026, GDP per capita grew 40%. The real minimum wage, over the same interval, grew 104%. The wage floor rose faster than the average for the economy.

The two measures are not exactly the same kind. The minimum wage is adjusted by the INPC; GDP is measured in volume, at constant prices. And GDP per capita includes income that never reaches anyone's wages, such as profits, rents and taxes.

The minimum wage and what working people earn

From 2012, it is possible to compare the minimum wage with what workers actually receive: that is when the PNAD Contínua, IBGE's survey on employment and income, begins. From March 2012 to August 2026, the real minimum wage rose 18% and the average real earnings of the employed rose 20%. They moved practically together. The minimum wage was equivalent to 43.6% of average earnings at the start of the series and 42.9% now. Along the way, that proportion fluctuated between 41.0%, in December 2014, and 48.9%, in January 2022.

Unemployment took its own path. The rate was 8.0% in the quarter ending in March 2012, reached 14.9% in September 2020, fell to the lowest level in the series, 5.1%, in December 2025, and stands at 5.3%. The number of employed people went from 87.7 million to 103.5 million.

Whether a higher minimum wage destroys jobs or not is an old debate among economists, with studies on both sides. These numbers do not settle the question: the minimum wage, employment and the whole economy moved at the same time, and the series alone cannot separate one thing from another.

Income inequality among households, measured by the Gini index, went from 0.540 in 2012 to 0.511 in 2025. The highest point was in 2018 (0.545); the lowest, in 2024 (0.504). Because the Gini adds up all incomes, including pensions and social programs, it does not isolate the effect of the minimum wage.

In dollars

In dollars, the story is different, because the exchange rate enters the calculation. The minimum wage was worth US$ 83 in January 1995, fell to US$ 53 in October 2002, reached US$ 362 in February 2012 and stands at US$ 315, with the dollar at R$ 5.14. The series does not discount U.S. inflation. It tells how much the minimum wage pays for dollar-priced goods, and is not a measure of purchasing power.

Interest rates, debt and expectations

The 2026 minimum wage coexists with double-digit interest rates. The Selic stands at 13.75% a year (Oct. 16, 2026), and the INPC accumulates 4.3% over 12 months. After inflation, the effective rate yields 9.8% a year. Gross government debt is at 82.9% of GDP (August 2026); in the series, it ranged from 51.3%, in December 2011, to 87.7%, in October 2020. Households owe the banks the equivalent of 49.9% of a year's income, versus 16.5% in January 2005, and delinquency reaches 5.0% of the credit portfolio. Two other analyses deal with these numbers in more detail: the one on household debt and the one on real interest rates.

In the Oct. 9 Focus survey, the median of analysts expects IPCA inflation of 5.22% in 2026 and 4.30% in 2027, GDP growth of 1.85% and 1.40% and a Selic of 13.25% and 12.00% at the end of each year. These are market projections, not Dado Bruto's.

Government by government

Splitting history by presidents is the most common way to compare the Brazilian economy, and also the easiest to distort. The table uses the same seven indicators for all periods, in chronological order and without ranking.

It has pitfalls. The January adjustment in the first year of each government was usually set by the previous government, but it counts toward the new one. The periods are not the same length: Dilma Rousseff's second term, cut short in May 2016, has 17 months; Lula's third, through September 2026, has 45. Some series start late, such as GDP per capita, in 2001, and the PNAD Contínua, in 2012. And each government faced different shocks, from the 2015–2016 recession to the pandemic. For the calculation, each period starts in the first full month; Michel Temer's, in June 2016.

No single column measures a government's performance on its own. Separating what was each president's decision from what came from the inherited rule or the economy requires checking the act behind each adjustment, a calculation left for another analysis.

Limits

The INPC measures inflation for households with income of one to five minimum wages, and each household's basket is different. The minimum wage is a national floor; state-level floors were left out. The average earnings in the PNAD Contínua include everyone from people who earn less than one minimum wage to people who earn many times more. The GDP and population series go through revisions. And the data end in different months: prices run through September 2026; earnings and unemployment, through August.

Nothing here measures cause and effect. Minimum wage, GDP, employment and earnings influence one another, and monthly or annual series do not allow one influence to be separated from the other.