On April 7, 2026, Trip.com Group-S (09961) announced an initiative. It randomly selected 6,000 employees from its headquarters, dividing them into an experimental group and a control group. The experimental group received an additional 45 days of discretionary leave per year—employees could take it at will without providing any reason or obtaining managerial approval. The control group continued under the existing policy. Performance metrics, turnover rates, and satisfaction levels for both groups were meticulously tracked and later submitted to an academic institution for independent analysis. This was not an employee benefits announcement; it was a randomized controlled trial, employing the same methodology Pfizer used to test its COVID-19 vaccine.
In the same week, Goldman Sachs released a report stating that AI is currently eliminating a net total of approximately 16,000 jobs per month in the United States, with Generation Z being the most affected. A lengthy article published by The New York Times just days prior suggested that the AI employment threat, once dismissed by many economists, is now being taken seriously by all. As machines begin to perform human tasks, should the very concept of "work" be redefined?
This article seeks to answer a simple question: What happened to companies over the past century that were bold enough to offer their employees more time off? The answer might surprise both employers and employees.
I. Henry Ford Was Not a Philanthropist In 1913, the annual employee turnover rate at Ford Motor Company's Highland Park factory ranged between 370% and 380%. To translate: for every worker Ford hired, the company had to rehire three to four people for the same position within a year. The assembly line was a recent innovation, and workers labored 10 to 16 hours daily for a daily wage of less than $2.50. The world's most efficient production line could not retain a single worker for long.
On January 5, 1914, Henry Ford announced two changes: doubling the daily wage to $5 and reducing the workday to 8 hours. Wall Street erupted. The New York Times described it as "blind generosity." Competitors declared Ford was digging his own grave—handing profits directly to workers was seen as akin to socialism. Ford ignored them. He had done the math. In the first year after the $5 day was announced, employee turnover plummeted from 370% to 16%. Absenteeism fell from 10% to below 0.5%. Recruitment and training costs dropped significantly. With a stable workforce and uninterrupted production lines, the cost to manufacture each Model T actually decreased.
This was just the initial calculation; the subsequent benefits were even greater. A $5 daily wage combined with an 8-hour workday meant workers finished their shifts by 3 or 4 PM. What did they do after work? They went shopping, consumed, and spent money—including on Ford automobiles. Workers built cars on the assembly line and then drove the cars they built back home. The producer and the consumer were unified in the same individual. This became known as "Fordism": mass production requires mass consumption, and workers must have both money and leisure time. Both are essential.
In 1926, Ford made another move: reducing the workweek from six days to five, with no reduction in weekly pay. Critics accused him of undermining the foundation of American industry. A decade later, the five-day workweek became the national standard, codified into law by the Fair Labor Standards Act of 1938. How did Ford himself explain it? In a 1926 public statement, he simply noted that shorter hours reduced fatigue-related errors, thereby boosting productivity. The language was crucial—no mention of "employee care" or "corporate social responsibility," only "increased productivity." It was purely the language of efficiency. Ford was always calculating, but he simply calculated further ahead than any of his competitors.
Viewed today, the choice Ford faced a century ago is strikingly similar to the dilemma in 2026: when technological leaps cause productivity to surge, you have three options: 1. Keep workers' hours the same, leading to overproduction and price collapse. 2. Lay off workers to cut costs, resulting in shrinking demand. 3. Reduce working hours and increase spending, keeping both production and consumption active. Ford chose the third option. The entire 20th-century American middle class began to grow from that decision. NPR's retrospective on that era concluded unequivocally: Henry Ford created the American middle class. The AI revolution of 2026 presents the same choice, but fewer are currently choosing the third path.
What happened? Ford became one of the greatest industrialists of the 20th century, and the five-day workweek became a global standard.
II. Iceland Didn't Collapse; UK Revenue Rose 35% In 1930, economist John Maynard Keynes predicted in his essay "Economic Possibilities for our Grandchildren" that by 2030, humans would need to work only 15 hours per week. With four years remaining until that deadline, his prediction is clearly far off the mark. However, Keynes was wrong not in direction, but in his estimation of human nature. He underestimated one thing: every time productivity gains free up time, managers tend to fill it with more work. Technology advances, but people don't work less. This pattern has held from the steam engine era through the AI age, unchanged for a century.
Fortunately, the 21st century has seen some data-driven initiatives. From 2015 to 2019, the Icelandic government conducted the world's largest trial of reduced working hours. 2,500 public sector and private employees participated, reducing their workweek from 40 hours to 35-36 hours, with no cut in pay. The think tank Autonomy released a report in 2021 summarizing the results as an "overwhelming success." Productivity remained the same or improved, and 86% of Iceland's workforce subsequently gained the legal right to request shorter hours. A 2024 follow-up report further confirmed that 97% of participants felt the shorter week improved or maintained their work-life balance, and 42% reported reduced stress. This was not a short-term honeymoon effect; the trial lasted five years. CNN's follow-up noted that Iceland's average annual GDP growth during the trial period was 4.1%, outperforming most other wealthy European nations.
Skeptics might say, "That's Iceland, a small country with a tiny population; it's not representative." Very well, consider the UK. In 2022, 61 companies with 2,900 employees adopted the "100-80-100 model": 100% of the pay, for 80% of the time, with a commitment to 100% of the output. The results? 92% of the companies chose to continue the policy after the trial ended. Average revenue increased by 35%. Sick leave decreased by 65%. 71% of employees reported reduced burnout. An NPR follow-up a year later found the positive effects persisted. By August 2025, The Guardian reported that over 200,000 UK workers had permanently switched to a four-day workweek. The transition from 61 companies to 200,000 people, from pilot to institutionalization, took just three years.
Microsoft Japan's 2019 trial was even more striking: giving employees five consecutive Fridays off, albeit for just one month. Productivity increased by 39.9%, electricity costs fell by 23%, and printing was reduced by 60%.
"Why do these numbers look so good?" Stanford economist John Pencavel provided a robust explanation. His econometric analysis of historical data from British munitions factories concluded that beyond 50 hours, output per hour declines sharply; after 55 hours, the output from additional hours is virtually zero. You might be sitting at your desk, but your brain has already clocked out.
OECD data paints a broader picture. Denmark's output per hour worked is $92.25, Germany's is $83.16, and France's is $81.56—among the highest globally. Meanwhile, the average annual working hours for a full-time employee in Germany is about 1,778 hours, significantly lower than in high-hour countries like South Korea and Mexico. Countries that work less are, in fact, wealthier.
It's akin to charging a phone. Going from 0% to 80% might take 30 minutes, but getting from 80% to 100% takes another 30 minutes. Human productivity follows a similar curve. The final 20% of working hours yield almost no output, yet they consume energy, contribute to depreciation, and accumulate fatigue that eventually leads to sick leave and turnover.
III. After Efficiency Gains: Layoffs or More Leave? All experimental data point to the same conclusion: working less leads to greater profitability for companies. Then, one looks at reality.
A study published in the Harvard Business Review in February 2026, conducted over eight months, observed a US tech company with about 200 employees after they voluntarily adopted AI tools. The result? Employees worked at a faster pace, their task scope expanded, and the boundary between work and home life blurred—sending prompts to AI during lunch, running "one more iteration" before bed. No one mandated this overtime; the employees did it themselves. One engineer stated: "You had thought that maybe you could work less. But then really, you don't work less. You just work the same amount or even more."
Data from ActivTrak, tracking 160,000 employees, provided a starker quantification: after AI adoption, weekend work increased by over 40%, email volume rose 104%, instant messaging surged 145%, and focused work time fell to a three-year low.
Where did the productivity gains from AI go? They did not translate into more time off. They translated into higher KPIs. Companies reaped the efficiency dividend and then assigned one person the work of ten, laying off the other nine. This logic is identical to that of 19th-century textile mills. More efficient spinning jennies were introduced, but adult workers didn't get to leave earlier—instead, child labor increased. It wasn't until Parliament legislated shorter hours that factory owners were surprised to discover that workers with shorter hours were more productive.
This pattern repeats: legislation comes first, followed by the discovery that "more time off is actually more profitable," while virtually no business leader proactively figures it out beforehand. A commentary in The Hill accurately described this structural dilemma: "Many CEOs publicly predict the imminent arrival of the four-day workweek, but none are willing to implement it first—quarterly earnings reports won't allow it." It's a classic prisoner's dilemma. Everyone knows shorter hours are beneficial long-term, but no one wants to bear the short-term pressure on quarterly results. Someone has to take the first step.
The story in China is particularly illustrative, and more optimistic than most assume. In August 2021, ByteDance abolished its "big/small week" system, where employees worked alternating six-day weeks. The employees' initial reaction was not joy, but calculation: overtime pay was eliminated, leading to an estimated 17-20% reduction in take-home pay. A detailed report by Geek Park noted that some discovered HR had initially factored the expected overtime pay into the total compensation package during salary negotiations. Public opinion was largely mocking, calling it a disguised pay cut.
But mockery aside, consider the data. After canceling the big/small week system, ByteDance did not "collapse." Its 2022 revenue reached approximately 550 billion RMB, a year-on-year increase of about 30%. Revenue grew to around $120 billion in 2023 and climbed further in 2024. Douyin's global daily active users continued to rise over these three years, and TikTok's commercialization expanded rapidly overseas. Abolishing the big/small week system neither slowed ByteDance's operational pace nor caused it to fall behind competitors. On the contrary, the years following the cancellation of mandatory overtime were among its fastest-growing periods.
In April 2025, Xiaohongshu followed suit. It canceled its big/small week system while simultaneously eliminating non-compete clauses for employees. Take-home pay was expected to drop 15-20%. A Xinhua Net commentary titled "Why is the Cancellation of 'Big/Small Week' Trending Again?" saw comment sections replaying the same script as ByteDance's case. However, Xiaohongshu's advertising revenue and e-commerce GMV maintained rapid growth during the same period, and its user base continued to expand. A Beijing Daily report defined this as "shedding the malaise of large internet companies."
Employees complained about reduced salaries. But employers should look at a different set of numbers: after both companies reduced the intensity of overwork, their business growth rates did not decline but actually increased. Overtime pay was cut, but profits were not. The 996 culture stopped, but growth continued.
Of course, the approaches of ByteDance and Xiaohongshu were not particularly refined. They eliminated mandatory overtime but did not redesign the work itself—the workload remained largely the same, only the overtime pay disappeared. The employees' perception was "less money, same amount of work." This falls short of "management innovation."
Trip.com Group-S (09961) took a different, more advanced path. In 2022, it implemented a "3+2" hybrid work model. James Liang, co-founder of Trip.com Group and a demographic economist, led a team that conducted a rigorous randomized controlled trial. The results were written into a paper published in Nature in 2024. The hybrid work group experienced a one-third reduction in attrition rates, increased job satisfaction, and saved the company millions in recruitment costs. This was published in Nature, not an internal corporate white paper. It took Trip.com four years to move from hybrid work to discretionary leave. Each step was backed by data. Randomly assigning 6,000 employees into groups, setting up control groups, collaborating with academic institutions—this is the level of experimental design typically seen in pharmaceutical clinical trials.
Three companies, three different approaches, but all validating one thing: being slightly less intense about overwork does not kill a company. ByteDance and Xiaohongshu demonstrated the lower bound in a crude way—eliminating mandatory overtime systems did not hinder breakneck growth. Trip.com is scientifically exploring the upper bound—can a company perform even better by giving employees greater autonomy?
Billionaire Mark Cuban predicted in an early 2026 interview: "Smart large companies will use AI Agents to boost efficiency, then reduce the workday by one hour, keeping salaries the same." He might be right. But he omitted half the sentence: smart large companies will dare to do this because others have tried it first, endured the initial criticism, and proven it works.
Even AI companies themselves are beginning to recognize this issue. In April 2026, OpenAI released a policy initiative titled "Industrial Policy for the Intelligence Age." One recommendation stood out: encourage companies and unions to pilot a four-day, 32-hour workweek with no reduction in pay or output, converting the saved time into permanently reduced hours or additional paid leave. They termed this "efficiency dividends." The fact that the company creating AI is explicitly stating that time saved by AI should be returned to workers is itself telling.
In 1926, Henry Ford cut the workweek from six days to five. Competitors said he was insane. A decade later, the five-day week was written into US federal law. In 1930, Keynes predicted a 15-hour workweek by 2030. Nearly a century later, we are far from that goal. The technology has long been capable. The obstacle has never been technology.
On April 7, 2026, Trip.com Group-S (09961) commenced its randomized controlled trial with 6,000 employees. Comment sections called it a publicity stunt. On Zhihu, users said, "waiting to see it fail." "The company will collapse if workers don't work"—this was Wall Street's assessment of Ford's $5 day in 1914. "The company will collapse if employees don't work overtime"—this is the modern executive's assessment of the four-day workweek in 2026. The identical sentiment, separated by 112 years.
So, what happened to companies that dared to give employees more time off? Ford became a titan of 20th-century industry. Iceland's GDP outperformed Europe's. UK pilot companies saw revenue rise by 35%. ByteDance entered a period of explosive growth. Xiaohongshu continued to expand its user base in the same year it canceled the big/small week. Trip.com Group-S published its experimental results in Nature.
A century of data, spanning five continents, involving tens of thousands of participants in experiments, all points in the same direction. But data rarely convinces anyone. Only profits can.
Comments