Formal labor market · Novo CAGED / MTE

CLT in Motion

Six years of formally registered jobs in Brazil, from the pandemic collapse to the 2025 slowdown, and what the forecasts and the debate over ending the 6×1 work schedule signal for formal employment through 2027.

10,006,260
net formal (CLT) jobs created since Jan/2020
1,211,827
net balance over the last 12 months
81%
of hires are contracted at 44h (the 6×1 universe)
166,654
balance projected for Jun/2027 (Seasonal Naive model)
Source: Novo CAGED microdata (PDET / Ministry of Labor and Employment) · Period Jan/2020–Mar/2026 · 27 states · Author's analysis · May 29, 2026
Overview

Today's snapshot

Brazil's formal employment is in a phase of decelerating expansion: it still adds jobs every month, but at a slower pace than at the peak of the post-pandemic rebound.

The recovery was real, but it ran out of steam. Between Jan/2020–Mar/2026, the cumulative balance reached 10,006,260 net formal jobs. Over the last 12 months the country created 1,211,827 formal jobs, versus 1,554,585 in the previous 12 months, a change of -22.0% that confirms the loss of traction amid high interest rates and costlier credit.

228,026
Latest monthly balance
Mar/2026
1,211,827
Balance · 12 months
hires − separations
24%
of the national balance
concentrated in São Paulo
0.69
Seasonal strength
marked annual pattern (0–1)
Trajectory

A six-year roller coaster

From the free fall of 2020 to the recent normalization: the monthly series tells the country's economic story better than any headline.

Monthly balance of formal (CLT) jobs in Brazil. Bars: monthly balance (green positive, red negative). Solid line: 12-month moving average. The dotted line before 2020 is the trend (12m MA) of the old CAGED (2007–2019), with 35 months from corrupt MTE archives filled by interpolation. The dashed line marks the Jan/2020 methodological break (Law 4.923/65 declarations → eSocial): the two eras are not strictly comparable. Source: old CAGED and Novo CAGED/MTE.

Apr/2020 is the bottom. In the first shock of the pandemic, the country shed -902,317 formal jobs in a single month, the worst result in the series. A V-shaped recovery followed, with record balances in 2021 and 2022 as the economy reopened.

Since then the moving average has settled into a positive plateau: Brazil keeps creating jobs, but momentum cooled in 2024–2026. The sawtooth pattern that recurs every year is not noise; it is seasonality, and it has a sectoral signature, as we will see.

Every December Brazil lays off; every start of the year, it rehires. The employment calendar is almost as predictable as the seasons.
Geography

Five Brazils on one map

Formal job creation is deeply concentrated. The Southeast accounts for about 41% of the recent balance, and São Paulo alone for 24%.

Cumulative 12-month balance, by region. Rolling sum, a read on the stock of jobs created over the past year in each region.

Explore each state's series and forecast:

Monthly balance and forecast to Jun/2027 by state (best model, 95% CI). Use the selector in the top-right corner.
Seasonality

The employment calendar

The average balance per month reveals an annual clock: the country hires in the first half and during harvests, and lays off in December. Retail and services set the rhythm.

Average balance by month of the year and state. Green = typical net creation; red = losses. The red December band is universal.

This pattern is dominated by labor-intensive sectors and full-time contracts, precisely those at the center of the debate over the 6×1 schedule. That is what we turn to next.

Forecast

Where we are heading

Five models, SARIMA, ETS (Holt-Winters), Seasonal Naive, Random Forest and LightGBM, project the series through June 2027. The one selected by validation was Seasonal Naive.

Brazil, recent history and forecasts to Jun/2027. Solid line: selected model; dashed: alternatives; band: 95% confidence interval.

The central reading is seasonal stability, not acceleration. For Jun/2027 the model projects a balance of 166,654 jobs, with a 95% interval between -200,211 and 533,519, a width that reflects the strong seasonality and the heteroskedasticity (unstable variance) detected in the series.

⚠️ The forecasts assume the current framework. A structural change such as ending the 6×1 schedule would be a break that these models, trained on the past, do not anticipate, which is why it deserves a chapter of its own.
Dossier · Working-time reform

Ending the 6×1 schedule

The proposal to replace the 6×1 schedule (six days of work, one of rest) with 5×2 is mobilizing Congress and the streets. What the CAGED microdata say about who would be affected, and what to expect.

What is at stake. The 6×1 schedule organizes the week into six workdays and one rest day, typically adding up to the 44 weekly hours allowed by the Constitution. The constitutional amendment to end 6×1 proposes capping the week at five days (5×2) and reducing hours, with no pay cut. It is the biggest debate over working time in the country since 1988.

Distribution of formal hires by contracted weekly hours (2020–2026). The highlighted bar, 44h, is the typical 6×1 universe.
81%
of hires at 44h
the legal maximum, the core of 6×1
88%
at 40h or more
full-time contracts predominate

Hiring data show that the 44-hour week is the norm, not the exception: about 81% of all formal hires since 2020 were agreed at the constitutional ceiling. Cutting the schedule therefore hits the modal contract of the Brazilian labor market.

But exposure is uneven across sectors. The map below crosses how dependent on a 44h week each sector is (horizontal axis) with how much employment it generates (vertical axis).

Map of sectoral exposure to the 6×1 schedule. Each bubble is a sector (CNAE); horizontal: % of hires at 44h; vertical: 12-month balance; size: hiring volume. Dashed line: national average of 81% at 44h.

The most exposed are also large employers. Sectors such as Agropecuária, Construção, Comércio e reparação and Indústria de transformação combine heavy dependence on the 44h week with significant job creation. Sectors above the average exposure account for about 64% of recent formal hiring; in other words, the reform falls precisely on the engine that hires the most.

CNAESector % at 44hBalance 12mHires 12m
AAgropecuária 96%-716 1,213,989
FConstrução 95%111,091 2,493,025
GComércio e reparação 90%226,533 6,214,111
CIndústria de transformação 89%75,465 3,680,512
HTransporte e armazenagem 87%94,338 1,454,659
EÁgua, esgoto e resíduos 87%14,168 161,557
LAtividades imobiliárias 85%5,585 93,904
IAlojamento e alimentação 84%82,647 1,684,840

What to expect, in light of the evidence

There is no consensus, and the final effect depends on the design (size of the cut, transition period, offsets). We cross the most-discussed scenarios with peer-reviewed economic literature (numbered at the end):

↑ Arguments in favor

Productivity per hour tends to rise when long shifts are cut: in empirical data output grows less than proportionally to hours1,2, and compressed schedules raise satisfaction and work attitudes4. There are also health and sleep gains from reducing long shifts7,8.

↓ Risks raised

It raises the hourly cost in labor-intensive sectors (retail, food service, services); employers tend to react by adjusting base pay and hours5. France's 35-hour experience did not produce a robust positive effect on employment3, suggesting caution about automatic job gains.

⟳ Likely adjustment

Reorganization of shifts, hour banks, more part-time shifts and 5×2 hires at 40h4,6. The burden would fall on the sectors in the right-hand quadrant of the map above, those with the greatest 44h exposure.

📊 For the model

A transition would change the level and seasonality of the series, a structural break. The forecasts in this report serve as a baseline (the “no reform” scenario) against which to measure the effect.

Academic evidence

References located by the paper-lookup scientific search protocol (k-dense scientific-agent-skills) via OpenAlex, with no fabrication: each item has a verifiable DOI.

  1. Pencavel, J. (2014). The Productivity of Working Hours. The Economic Journal. · doi:10.1111/ecoj.12166
  2. Collewet, M.; Sauermann, J. (2017). Working hours and productivity. Labour Economics. · doi:10.1016/j.labeco.2017.03.006
  3. Chemin, M.; Wasmer, E. (2009). Using Alsace-Moselle Local Laws to Build a Difference-in-Differences Estimation Strategy of the Employment Effects of the 35-Hour Workweek Regulation in France. Journal of Labor Economics. · doi:10.1086/605426
  4. Baltes, B. B. et al. (1999). Flexible and compressed workweek schedules: A meta-analysis of their effects on work-related criteria. Journal of Applied Psychology. · doi:10.1037/0021-9010.84.4.496
  5. Trejo, S. J. (1991). The Effects of Overtime Pay Regulation on Worker Compensation. American Economic Review. · (no DOI)
  6. Deakin, S.; Wilkinson, F. (1988). Working Time and Employment. The Economic Journal. · doi:10.2307/2233396
  7. Afonso, P.; Fonseca, M.; Pires, J. F. (2017). Impact of working hours on sleep and mental health. Occupational Medicine. · doi:10.1093/occmed/kqx054
  8. Harrington, J. M. (2001). Health effects of shift work and extended hours of work. Occupational and Environmental Medicine. · doi:10.1136/oem.58.1.68

An economic and exploratory analysis based on the composition of formal hiring. CAGED records contracted hours, not the number of days worked; 44h is used as a proxy for the 6×1 universe. The cited evidence comes from institutional contexts different from Brazil's and should not be read as a forecast of the amendment's effect, nor as advice.

Appendix

State-by-state summary

Recent balance, selected model, forecast for Jun/2027 with confidence interval and diagnostics by state. Click the headers to sort.

StateName Balance 12mModelForecast Jun/2795% CI Seas.Heter.
RORondônia 7,199SARIMA 984 -1,981 a 3,949 0.67 yes
ACAcre 4,289Seasonal Naive 605 -364 a 1,574 0.80 no
AMAmazonas 18,520SARIMA 3,735 -1,054 a 8,524 0.58 yes
RRRoraima 1,402Random Forest 226 -2,681 a 3,133 0.61 no
PAPará 30,876SARIMA 6,258 -1,308 a 13,825 0.85 yes
APAmapá 6,042LightGBM 1,123 -1,598 a 3,845 0.69 yes
TOTocantins 4,731Seasonal Naive 513 -1,368 a 2,394 0.84 yes
MAMaranhão 30,316Seasonal Naive 6,247 1,867 a 10,627 0.78 yes
PIPiauí 20,923Random Forest 2,746 -5,342 a 10,834 0.73 no
CECeará 55,335Seasonal Naive 7,320 -4,947 a 19,587 0.70 no
RNRio Grande do Norte 16,184Random Forest 2,278 -10,166 a 14,722 0.80 yes
PBParaíba 28,390SARIMA 2,434 -1,644 a 6,511 0.80 no
PEPernambuco 73,554Random Forest 6,316 -30,150 a 42,782 0.82 yes
ALAlagoas 16,347SARIMA 2,394 -3,286 a 8,075 0.86 yes
SESergipe 18,526Seasonal Naive 2,407 -485 a 5,299 0.82 yes
BABahia 87,732Random Forest 9,207 -40,334 a 58,749 0.52 yes
MGMinas Gerais 72,941SARIMA 28,572 -21,987 a 79,132 0.74 yes
ESEspírito Santo 18,230Seasonal Naive -3,348 -12,089 a 5,393 0.65 yes
RJRio de Janeiro 113,440Random Forest 12,776 -86,566 a 112,119 0.44 yes
SPSão Paulo 288,486SARIMA 49,375 -80,240 a 178,991 0.64 yes
PRParaná 75,469Seasonal Naive 9,377 -15,893 a 34,647 0.75 yes
SCSanta Catarina 52,795SARIMA 8,199 -18,353 a 34,751 0.68 yes
RSRio Grande do Sul 28,211Random Forest 584 -62,261 a 63,428 0.67 yes
MSMato Grosso do Sul 20,922Seasonal Naive 2,709 -2,372 a 7,790 0.84 yes
MTMato Grosso 27,718Seasonal Naive 9,388 1,939 a 16,837 0.96 yes
GOGoiás 43,169SARIMA 8,033 -8,038 a 24,103 0.77 yes
DFDistrito Federal 50,080SARIMA 5,833 -2,102 a 13,769 0.69 yes