The four-day week: what the evidence actually shows

Reduction or compression: two models not to be confused

Working-time reduction means cutting the number of hours worked while pay stays the same. This is the model tested in the large international pilots under the "100-80-100" formula: 100% of pay, 80% of the time, 100% of the expected output. The equation only holds if the company genuinely removes work — shorter meetings, fewer interruptions, automation, hard choices about low-value tasks. Without that groundwork, reduction turns mechanically into intensification.

Compression keeps the hours intact and redistributes them across four days. An employee on 35 hours moves to 8 hours 45 minutes a day; one on 39 hours moves to 9 hours 45. The gain is real — an extra free day — but it is paid for in heavier days, end-of-day fatigue and childcare constraints. French unions see it as a false advance: the CGT, France's largest confrontational union and a long-standing advocate of a 32-hour week, explicitly distinguishes the "four-day week" (with reduction) from the "week in four days" (without), which it calls a step backwards whenever the workload is left untouched.

A third, quieter variant also circulates: moving to four days with a proportional pay cut. That is neither model — it is 80% part-time work, which has existed in French law for decades. So when a job ad announces a "four-day week", the first questions to ask are about hours and pay.

What French labor law allows — and requires

There is no legal framework specific to the four-day week in France. Both models are built out of existing provisions of the labor code (code du travail), and the constraints differ depending on which model you choose.

The statutory working week remains 35 hours (article L3121-27). That is not a ceiling but the threshold above which overtime kicks in. Nothing requires those 35 hours to be spread over five days.

Compression is bounded by the maximum working times. Actual daily working time may not exceed 10 hours (L3121-18), raised to a maximum of 12 hours by collective agreement in periods of increased activity or for reasons linked to work organization. Weekly hours are capped at 48 (L3121-20) and at an average of 44 over twelve consecutive weeks (L3121-22), an average that may be raised to 46 hours by agreement (L3121-23). On top of that come 11 hours of daily rest and 35 consecutive hours of weekly rest. In practice, 39 hours over four days fits comfortably; 44 hours no longer does.

Cutting to 32 hours paid as 35 requires an agreement. Changing the working time written into the employment contract requires either a company-level collective agreement (accord d'entreprise) or an addendum (avenant) signed by each employee. This is the main practical difference from compression: how hours are scheduled falls under the employer's managerial authority unless the contract says otherwise, whereas a reduction in contractual hours cannot be imposed unilaterally.

The forfait jours follows its own logic. This annualized-days contract, reserved for autonomous managerial staff and employees whose working time cannot be set in advance (L3121-58), counts days worked rather than hours, up to a limit of 218 days a year set by collective agreement (L3121-64). Moving a forfait jours to four days a week means reopening the agreement and revisiting the number of days, along with the workload monitoring and right-to-disconnect provisions on which its validity depends.

What the trials abroad measured

Iceland (2015-2019). Two trials run by the City of Reykjavík and the Icelandic state covered 66 workplaces and around 2,500 workers, more than 1% of the country's working population. Hours fell from 40 to 35-36 with no loss of pay. The evaluation by Autonomy and Alda concludes that productivity was maintained or improved, well-being clearly improved and the exercise was budget-neutral, and states that 86% of Iceland's working population has since obtained shorter hours or the right to shorten them. That reading is disputed: The Conversation points out that this was not a four-day week, that the largest cut amounted to four hours a week and was limited to one to three hours in 61 of the 66 workplaces, and that the national agreements which followed granted only 35 minutes a week in the private sector. The trials had no control group.

United Kingdom (2022). The pilot coordinated by 4 Day Week Global with Autonomy and researchers from Boston College and Cambridge brought together 61 companies and around 2,900 employees from June to December 2022, on the 100-80-100 model. Resignations fell 57% over the period. Revenue was broadly stable, up 1.4% on average when weighted by company size — the widely quoted "+35%" figure is a comparison with the same period a year earlier, not the effect of the scheme. At the end of the trial, 56 of the 61 companies (92%) carried on. A year later, at least 54 still had it in place and 31 had made it permanent.

Portugal (2023). The pilot backed by the Portuguese government and evaluated by Henley Business School and Birkbeck covered 41 companies, 21 of them in a coordinated six-month trial launched in June 2023, and more than 1,000 employees. Actual weekly hours fell from 41.6 to 36.5. The share of employees struggling with work-life balance dropped from 50% to 16%, fatigue fell by 23% and sleep problems by 19%. Eighty percent of managers judged the exercise financially neutral, and 37 of the 41 companies kept the arrangement.

Germany (2024). Forty-five companies took part in a six-month pilot organized by the consultancy Intraprenör and 4 Day Week Global, with the University of Münster. About 40% confined it to certain teams. Objective measurements show, relative to the control group, 38 more minutes of sleep per week and nearly 1,850 additional steps per week. At follow-up, 70% of the organizations were continuing with some form of working-time reduction, 22% of them after altering their original model.

The most robust study to date was published in 2025 in Nature Human Behaviour by Wen Fan and Juliet Schor: 141 organizations, 2,896 employees, six English-speaking countries, with 12 control organizations. The actual reduction observed is around five hours a week. Work-related burnout falls from 2.83 to 2.38 on a scale of 1 to 5, and job satisfaction rises from 7.07 to 7.59 out of 10, driven by better sleep, less fatigue and greater work ability. The authors flag the limits themselves: the companies volunteered and are therefore not representative, there was no randomization, the data are self-reported, and the sample is dominated by professional services.

What actually happened at LDLC

LDLC, a computer hardware retailer based in the Lyon area, moved to a four-day week in January 2021 for its entire workforce of roughly a thousand people. The model chosen is reduction: 32 hours a week instead of 35, with no pay cut, spread over four eight-hour days — the company ruled out nine-hour days as too heavy for parents. Managerial staff on a forfait jours switched to a version of the contract without RTT days (the compensatory days off French employees accrue when they work beyond 35 hours). The initial extra cost was estimated at around 5% of payroll, and pay rises were nevertheless negotiated the following year.

The company reports absenteeism and workplace accidents cut in half, turnover down to almost nothing, and a sharp rise in unsolicited job applications. The much-repeated "+40% productivity" figure needs qualifying: founder Laurent Villemonte de la Clergerie explains it by an increase in activity of roughly 40% during the Covid period, absorbed with no net hiring, headcount staying at around 1,000. So it reflects apparent productivity over an exceptional period for the computer market, not a gain measured all else being equal.

Two caveats apply. No independent academic evaluation has been carried out on the LDLC case: the public figures come from the company and its chief executive, and vary from one interview to the next. And an online retailer, whose operations lend themselves to automation and to reorganizing workflows, cannot be transposed as-is to a hospital or a factory running three shifts around the clock.

Where France stands

Public opinion is overwhelmingly in favor — of one specific model. An Ifop poll for Politis, conducted on 12 and 13 March 2024 among 1,008 people representative of the population aged 18 and over, asked explicitly about "the four-day week, 32 hours paid as 35". Seventy percent of French respondents are in favor, rising to 77% of working people, 84% of under-35s and 81% of blue-collar workers — and 72% of the business leaders surveyed. By contrast, only 49% of those aged 65 and over are in favor.

Actual working hours remain long. According to Insee, the national statistics institute, a full-time employee reported usual weekly hours of 38.8 in 2024, for 1,661 hours actually worked over the year; managerial staff are at 42 hours and 1,781 hours. The French debate is therefore not about moving from a theoretical 35 hours to 32, but about a far more substantial cut in practice.

Pioneering employers remain few, and self-selected. Besides LDLC, the Lyon-based company Elmy moved in September 2022 to 32 hours for non-managerial employees and 35 hours for managerial staff with no pay cut, an arrangement made permanent after a six-month trial, with absenteeism in its customer service team cut by four according to the company. The materials recycler Yprema has been gradually reducing its teams' working hours since the late 1990s. A first collective French experiment was launched in 2025 by 4jours.work with emlyon business school, B-Corp France and the CFE-CGC union across some twenty volunteer organizations, extended by a second phase in February 2026; the results are too recent to be treated as established.

The state, for its part, is testing compression. A DGAFP circular of 22 March 2024 launched an experiment in the state civil service, starting in spring 2024, on a voluntary basis and subject to the approval of the head of service. The annual total of 1,607 hours stays unchanged: the point is to concentrate it into four days, meaning roughly nine hours a day. Teachers and staff on special shift cycles are excluded.

Belgium and Spain: two ways to fail by legislating

Belgium legislated, and almost nobody took it up. The law of 3 October 2022, in force since 20 November 2022 as part of the "Labour Deal" (Deal pour l'emploi), gives every private-sector employee the right to request working their full-time hours over four days. The employer must give written reasons for any refusal, and the employee making the request is protected against dismissal. But the scheme is pure compression: a Belgian full-time week of 38 hours becomes four days of nine and a half hours. At the end of July 2024, an analysis by the payroll agency Acerta covering 21,000 employers and 220,000 full-time employees on permanent contracts found that 0.75% of private-sector employees were using it, across 2.75% of companies. The rate rises to 36% among employers with more than 500 staff and falls below 6% in companies with fewer than 50. The request must also be revalidated every six months.

Spain tried both routes, without decisive success. A Ministry of Industry pilot launched in 2023 put €10 million on the table to help industrial SMEs cut working time with no loss of pay, in exchange for a productivity-improvement plan. On the legislative front, labor minister Yolanda Díaz's bill lowering statutory hours from 40 to 37.5 a week was rejected by the Congress of Deputies on 10 September 2025, by 178 votes to 170.

The serious objections

Work intensification is the central risk. Cutting hours without cutting the workload means asking for the same output in less time. The systematic review of the literature on the subject identifies intensification as the most frequently documented adverse effect, together with tighter performance measurement and monitoring. It is also the French unions' core criticism: without negotiation over workload, the reduction is paid for by employees themselves.

Compression does not deliver the expected health benefits. A longitudinal study at IKEA Belgium covering 204 employees, 140 of whom moved to compressed schedules in 2024, published in BMC Public Health, shows that psychological detachment from work improves — employees switch off better on their days off — but that burnout risk does not fall. The extra day works as a "micro-vacation" whose effect fades on return, because the stressors are left intact.

The available evidence suffers from selection bias. The companies that sign up for a pilot are the ones that already believe it can work for them, and the decision comes from management. That bias weighs more heavily on the economic results than on the well-being indicators. Add to it the self-reporting by employees — who have an obvious interest in keeping their free day — and the scarcity of control groups: the Nature Human Behaviour study has only 12 for 141 organizations.

The cost of generalizing it remains poorly assessed. In activities where presence is the service, shorter hours cannot be absorbed through efficiency gains: they require hiring, with no guaranteed offset. No existing pilot allows this question to be settled at the scale of an economy, since all of them cover volunteer companies, mostly in services.

The sectors where it breaks down

Continuous-service activities are structurally resistant to the model. An emergency department, a care home, a production line running three shifts, a shop open six days a week cannot "do the same thing in less time": the time is the service delivered. Any reduction there translates into a need for extra staff, in occupations that already struggle to recruit.

Very small businesses run into a different obstacle. The Belgian experience is unambiguous: fewer than 6% of employers with under 50 staff offer the arrangement, against 36% of those above 500. When every single person is indispensable at all times, closing for a day or organizing a rotation becomes unworkable.

Finally, physically demanding jobs cope badly with compression: stretching the day from 8 to 10 hours increases exposure to risk instead of reducing it. In those roles, only genuine working-time reduction is defensible.

How a company actually goes about it

The pilots that succeed share the same sequence, and it starts a long way from questions of scheduling. The 2022 UK pilot allowed two months of preparation before any change of rhythm: workshops, mentoring, exchanges between participating companies. The purpose of that phase is to identify what the organization is going to stop doing.

The recurring levers are few: cut the number and length of meetings, protect blocks of uninterrupted work, clarify decision-making paths, automate what can be automated, make hard calls on low-value tasks. That is the work that produces the hours — not the decision to open up a day off.

The format comes next: a single common day off, rotating days to maintain service continuity, or a nine-day fortnight — in Portugal, companies split almost evenly between the weekly day and the fortnight. That leaves the legal side — collective agreement or individual addenda, revising forfait jours arrangements, checking the maximum working times — and the definition of indicators measured before and after: absenteeism, turnover, processing times, customer satisfaction. Without a baseline, no assessment will stand up to challenge.