Shorter working weeks in Brazil: a model separates 40-hour and 36-hour limits

A productivity requirement, not a forecast

How much more efficiently would Brazil’s economy need to operate to maintain output after a shorter working week? Victor Rangel and Fernando Barros Jr address that question in an arXiv preprint, rather than estimating what a particular law would actually cause.

Their central calculation is a break-even requirement: the increase in total factor productivity, or TFP, that would restore output after introducing a lower weekly-hours ceiling. TFP describes how effectively capital and labor are combined to produce output. It is not simply output per hour, and the calculation does not assume that shorter hours will automatically deliver the required improvement.

Under the study’s main assumptions, a 40-hour ceiling requires a one-time TFP increase of 1.53–1.57%. A 36-hour ceiling requires 6.52–8.38%. Those ranges reflect alternative model assumptions, not statistical confidence intervals.

How the calculation works

The researchers use Brazil’s PNAD Contínua household survey for the fourth quarter of 2024. Their weighted sample includes 209,311 employed people aged 14 or older, with information about their main job. Informal employment accounts for 38.60% of employment in their estimates.

They feed observed working-hour distributions, informality and relative hourly earnings into a static, partial-equilibrium model. Here, “static” means it compares situations without tracing an adjustment process over time. “Partial equilibrium” means it leaves wider economic feedbacks outside the calculation.

Capital and total employment remain fixed. Firms can change their mix of formal and informal workers, but these two labor inputs cannot replace each other perfectly. Costs associated with formalization, informality and changing the employment mix influence firms’ choices. The model therefore permits employment to move between categories, but cannot estimate job creation or unemployment effects.

Hourly efficiency depends on working time. Crucially, the authors assume its peak occurs at 40 hours; they do not estimate a Brazilian optimum from these data. One specification penalizes efficiency only above 40 hours. Another also penalizes hours below that point. Existing research motivates fatigue effects, but does not identify these particular curves for Brazil.

Why the two ceilings differ

Moving toward 40 hours removes some hours that the model treats as fatigue-reducing opportunities. Moving below 40 removes hours without further efficiency gains in one specification, and introduces an efficiency penalty in the other.

Without compensating TFP gains, modeled output falls by about 1.64–1.68% at 40 hours, compared with 6.65–8.40% at 36 hours. The productivity requirements are calculated separately, allowing firms to readjust their employment mix as productivity changes.

At 36 hours, informality increases by 2.28–2.92 percentage points from its starting level. Nevertheless, the researchers’ decomposition attributes the largest negative output contribution to fewer physical hours in formal employment, rather than movement into informality. Efficiency either cushions or worsens the loss, depending on the assumed curve.

The starting-hours assumption matters substantially

The headline results first truncate formal workers’ reported usual hours at 44 hours. This reduces their starting average from about 41.42 to 39.95 hours. It is a measurement assumption designed to represent a 44-hour baseline—not evidence about actual contractual hours or how a new law would apply to overtime.

That choice matters because 17.97% of formal workers report usual hours above 44. When the researchers retain this longer-hours tail and rebuild the calibration, the productivity requirement at 40 hours falls to approximately 0.07%. At 36 hours it becomes 5.00–6.80%.

Removing fatigue effects altogether gives requirements of 2.40% at 40 hours and 7.42% at 36 hours under the capped baseline. The results therefore depend materially on both the assumed productivity response to hours and the starting distribution.

Output is not the whole welfare calculation

The paper also weighs consumption against time outside paid work using a representative-worker preference model. Under the central calibration, its consumption-equivalent welfare measure improves slightly at 40 hours, but declines at 36 hours when productivity is unchanged.

This is a limited diagnostic, not a complete social ranking. It excludes health, household production, bargaining, distributional differences and employment transitions. Alternative starting-hours and fatigue assumptions can also change its conclusions.

The model additionally omits investment responses and economy-wide wage, price and sectoral feedbacks. Several important parameters are assumed rather than independently identified from Brazilian evidence. The preprint thus offers conditional benchmarks for comparing reform intensity—not proof that either ceiling is desirable, infeasible or certain to produce the simulated losses. Gradual implementation warrants investigation, but the benefits of a transition timetable would require a dynamic analysis that this study does not provide.