Skip to content
Sign inGet started free

The record · 11 September 2026

In a year, Brazil gained 335,000 people working for themselves with a registered business.The number working for themselves did not measurably grow.

One release, read down the column most readers skip. Every figure below is IBGE's own, including its verdict on which changes the survey can vouch for. Nothing here is a cause, and everything is checkable in the published tables.

· 8 min read

+335,000people working for themselves with a registered business (CNPJ), May to July 2026 against a year earlier, a change IBGE marks statistically significant
+162,000people working for themselves in total, registered or not, a change inside the survey's margin of error

Both are changes on a year earlier. Only the first clears the margin.

5.3%unemployment, May to July 2026: the lowest for those three months since the series began in 2012, and not the lowest reading ever, which was 5.1% at the end of 2025

The column most readers skip

IBGE's survey for May to July 2026, published on 27 August, is a good report by the measures it is usually read by. 103.3 million people in work, 899,000 more than a year earlier. Unemployment at 5.3%, down from 5.6%. Real average earnings of 3,762 reais a month, up 3.3%. All three changes clear the survey's margin of error, and IBGE says so.

The 5.3% is the lowest reading for May to July since the series began in 2012, and that is the precise claim. It is not the lowest reading ever: the three months to December 2025 came in at 5.1%. Earnings are not at a peak either; they stood at 3,805 reais in January to March 2026.

What the headlines leave out is the column beside each number. For every line of the release, IBGE states whether the change on a year earlier is statistically significant at 95% confidence, or whether it could be the sampling of a household survey. Of the twelve lines below, eight clear that bar and four do not, and the four are where the reading changes.

Provenance: Source
People aged 14 and over, in thousands, May to July 2025 and May to July 2026, with the change and IBGE's own verdict on it at 95% confidence. The twelve lines this piece discusses, in the order IBGE publishes them. IBGE computes each change before rounding, so subtracting the two columns can differ from it by one thousand.
LineMay to Jul 2025May to Jul 2026ChangeIBGE verdict
In work102,437103,335+899Significant
Unemployed6,1185,819-299Significant
Outside the labour force65,63566,356+721Significant
Private-sector employees52,58353,180+597Significant
of which, with a signed work card39,11339,428+314Within the margin
of which, without one13,47013,753+283Within the margin
Domestic workers5,6455,409-236Significant
Public-sector employees without a work card, civil servants and military excluded3,2693,507+238Significant
Employers with a CNPJ3,3603,595+235Significant
Working for themselves25,93826,099+162Within the margin
of which, with a CNPJ6,9407,276+335Significant
of which, without one18,99818,824-174Within the margin

Registered is not hired

People working for themselves with a CNPJ rose from 6.94 million to 7.28 million, a change of 335,000 that IBGE marks significant. People working for themselves in total, with or without one, rose by 162,000, and that change is inside the margin: the survey cannot tell it from no change at all. Those without a CNPJ fell by 174,000, also inside the margin.

Employers moved the same way. Employers with a CNPJ rose by 235,000, significant; employers without one did not move measurably. Among the people who run a business, alone or with staff, the clear change is in the registered half.

A CNPJ is an entry in Brazil's national register of legal entities. It is not the signed work card, the carteira assinada, which is what makes someone an employee on a formal payroll. The survey counts the two separately, and so does this piece.

Provenance: SourceThree terms the release uses exactly
  1. Significant, or within the margin. A household survey measures a sample, so every change carries an error. IBGE marks a change significant only when it clears 95% confidence. Within the margin does not mean nothing happened; it means the survey cannot tell.
  2. Registered: a CNPJ. The Cadastro Nacional da Pessoa Jurídica, Brazil's national register of legal entities. The survey asks whether a business, including one person working alone, is in it.
  3. On the books: a signed work card. The carteira assinada is what makes an employee formal. It belongs to employees, and a CNPJ is not one.

So "335,000 more registered" and "335,000 more hired" are different sentences, and only the first is in the release. None of the three terms says why a number moved.

The finding rests on three distinctions IBGE draws in its own tables. A reading that blurs any one of them says something the survey does not.

What the margin will not let anyone say

It will not say which kind of job the private sector added. Private-sector employment rose by 597,000, and IBGE marks that significant. Its two halves, with a signed work card (+314,000) and without one (+283,000), are each inside the margin. The survey is sure the total grew and cannot say in which half. A reading of this release as a formal-jobs boom, or as an informal one, claims more than the data holds.

It will not say informality fell. The quarterly informality rate for April to June was 37.4%, against 37.8% a year earlier, and IBGE marks that change not significant.

Two movements are clear and point elsewhere. Domestic work fell by 236,000. Public-sector employees without a work card, outside the civil service and the military, rose by 238,000. And the population aged 14 and over grew by 1.32 million, of whom 600,000 more are in the labour force and 721,000 more are outside it, both significant.

Nor will it say who moved where. The survey compares two samples a year apart; it does not follow the same people. So it cannot show that anyone left a payroll for a CNPJ, or a CNPJ for a payroll, and this page does not claim that either happened.

What the data establishes

  • In IBGE's survey for May to July 2026, 103.3 million people were in work, 899,000 more than a year earlier, and unemployment was 5.3%, down from 5.6%. IBGE marks both changes significant.
  • People working for themselves with a CNPJ rose by 335,000, to 7.28 million, and employers with a CNPJ by 235,000. IBGE marks both changes significant.
  • People working for themselves in total rose by 162,000, a change within the margin, and those without a CNPJ fell by 174,000, also within it.
  • Private-sector employment rose by 597,000, significant; its halves with a signed work card (+314,000) and without one (+283,000) are each within the margin.
  • 5.3% is the lowest May-to-July reading since the series began in 2012. The lowest reading in the whole series is 5.1%, for October to December 2025.

What it does not

  • That anyone swapped a payroll job for a CNPJ. The survey compares two samples and follows no one; a shift between groups is not a movement of people.
  • Any cause for the rise in registrations, whether tax rules, employers taking on contractors instead of staff, or anything else. Each is argued in Brazil and none is measured here.
  • That formal jobs grew, or that informal ones did. The private sector grew and the survey cannot place the growth in either half; the informality rate did not change significantly.
  • That unemployment is at a record low. It is at its lowest for May to July; the series low is 5.1%.
  • That a person with a CNPJ works on the terms of a signed work card. The survey counts them separately because they are different things.
  • Any trend beyond this comparison. Two moving quarters a year apart are compared here, and nothing else.

Method

  1. One release. IBGE, Pesquisa Nacional por Amostra de Domicílios Contínua (PNAD Contínua), moving quarter May to July 2026, published 27 August 2026, read from IBGE's SIDRA tables 6318 (labour force status), 6320 (position in occupation), 6381 (unemployment rate) and 6390 (real average earnings). The informality rate is quarterly: table 8529, April to June 2026, published 14 August 2026.
  2. THE VERDICT IS IBGE'S, NOT OURS. Each table carries, beside every change on the same moving quarter a year earlier, a status IBGE computes from its own sampling errors: significant at 95% confidence, or not. The desk did not recompute a single interval. Where this page says within the margin, it is repeating IBGE.
  3. ROUNDING. Levels are in thousands as IBGE publishes them, and IBGE computes each change before rounding. So 102,437 to 103,335 subtracts to 898 on the page, while IBGE's change is 899. The table prints the levels and IBGE's change as published rather than forcing them to agree.
  4. ONE BASIS. On 31 July 2025 IBGE moved the survey to a new weighting and revised its historical series (Nota Técnica 02/2025). Every figure here, 2025 and 2026 alike, comes from the revised series as published today, so the two years stand on the same footing.
  5. THE SUPERLATIVE WAS CHECKED ON THE WHOLE SERIES. 5.3% was compared with all fifteen May-to-July readings since 2012, the lowest before it being 5.6% in 2025, and with all 173 moving quarters, whose lowest is 5.1% (October to December 2025). The same check on earnings found a higher reading, 3,805 reais, in January to March 2026.
  6. WHAT A CNPJ LINE MEANS. It is the survey's own question: whether the business of a person who works for themselves, or who employs others, is registered in the CNPJ. This page reads nothing further into the registration, neither its type nor its reason.

BabZituna scores every job against your profile on six real dimensions, and publishes what it can and cannot measure. See the live market.

The markets in this piece, live

Earlier editions