Long Working Hours and Productivity: The 50-Hour Wall for CEOs

Past about 50 hours a week, extra hours stop buying much extra output. The best-known study on it found that output at 70 hours a week was barely different from output at 56, and the health data shows stroke risk rising once people regularly work 55 hours or more.

Key takeaways

Key takeaways

1. In Pencavel's 1916 munitions data, output tracked hours up to about 49 a week, then flattened; 70 hours produced little more than 56.
2. Losing the weekly rest day cost about 10% of output at the same weekly hours.
3. Working 55 or more hours a week is linked to roughly 35% higher stroke risk (WHO/ILO 2021) and 1.33 times stroke risk in a 600,000-person Lancet meta-analysis, as associations rather than proof.
4. In 2023, 26.0% of self-employed Canadians worked 49 or more hours a week, against 5.5% of employees.
5. A practical target for a CEO is a 50-hour budget, measured by a two-week honest log, with one full day at zero work.
6. Present the change to a board as risk management with a 90-day trial and agreed measures.

So my answer to "how many hours should a CEO work" is a budget of about 50, with one full day off, and a log that proves where the hours are going. Below is the evidence, what it can and cannot tell you, and the system I would set up.

Do long working hours actually increase productivity?

Up to a point, yes. After that point, much less than you think.

The study everyone cites is by Stanford economist John Pencavel, published as an IZA discussion paper in 2014. He went back to data collected by Britain's Health of Munition Workers Committee during the First World War, mostly in 1916. The main dataset is 122 weekly observations on four groups of munitions workers (mostly women) in state-controlled factories, with weekly hours ranging from 24 to 72.5.

What he found:

• Below about 49 hours a week, output rose in step with hours. Work 10% more, make roughly 10% more.
• Above 49 hours, output kept rising but at a falling rate. In his words, "Output at 70 hours differs little from output at 56 hours."
• Working without a weekly rest day cost about 10% of output, even holding total weekly hours the same.
• A 48-hour, six-day week produced slightly more than a 70-hour, seven-day week: about 1.4% more.

Read that last line again. Twenty-two extra hours and a lost Sunday, for slightly less shell production.

What the 1916 munitions data cannot tell you

I like this study, and I also think founders lean on it too hard. Pencavel is upfront about the limits. The sample is small enough that he says the work "may appear almost anecdotal." It was wartime, so people may have pushed through fatigue they would not tolerate in peacetime. The workers in the data were the ones healthy enough to show up, which probably understates the fatigue effect. And the job was physical, repetitive factory output, measured in units.

Your job is judgment: hiring, capital, priorities, saying no. Nobody has a clean output curve for that. My read is that judgment work degrades at least as fast as machine work when you are tired, but that is my opinion, not a finding. What the munitions data does give you is a strong prior: past a certain weekly load, more hours are a poor trade.

What does a 55-hour work week do to your health?

In 2021 the World Health Organization and the International Labour Organization published the first global estimate of deaths linked to long hours. Their figures, for 2016:

• 745,000 deaths from stroke and ischemic heart disease were attributed to working 55 or more hours a week, up 29% since 2000.
• That split into 398,000 stroke deaths and 347,000 heart disease deaths.
• About 488 million people, or 8.9% of the world's population, were working 55-plus hours.
• Compared with a 35 to 40 hour week, working 55 or more was linked to an estimated 35% higher risk of stroke and 17% higher risk of dying from ischemic heart disease.
• 72% of the deaths were in men, and most were people aged 60 to 79 who had worked 55-plus hours between the ages of 45 and 74.

That last point matters for the founder in their 40s. The damage shows up later than the hours do.

The second big source is a 2015 meta-analysis in The Lancet led by Mika Kivimäki. It pooled 25 studies from 24 cohorts in Europe, the US and Australia: 603,838 people for coronary heart disease and 528,908 for stroke. People working 55 or more hours had a 1.33 times higher risk of stroke than people working 35 to 40 hours, and a 1.13 times higher risk of coronary heart disease. The stroke risk climbed in steps as hours rose, which is what you would expect if hours themselves were part of the story.

Association, not proof

These are observational studies. They show long hours travel with higher risk; they do not prove the hours cause it. The WHO/ILO authors say so themselves: they cannot rule out other factors driving the link, working hours were mostly self-reported, and the evidence was rated moderate quality. Long hours tend to come bundled with less sleep, less movement, more stress and skipped checkups, and the studies cannot fully separate those.

From a coaching point of view, that bundle is the point. You do not have to settle the causation argument to see that a 60-hour week leaves almost no room for the sleep, training and recovery that keep a leader sharp. This is general information, not medical advice, and if you have heart or blood pressure concerns, start with your doctor.

The evidence at a glance

How many hours do CEOs and founders actually work?

More than the threshold, almost always.

Harvard's Michael Porter and Nitin Nohria tracked 27 CEOs of large companies (average revenue US$13.1 billion) around the clock in 15-minute blocks for 13 weeks each, about 60,000 hours of data in total, and published it in Harvard Business Review in 2018. As reported, the CEOs averaged 9.7 hours on weekdays and 62.5 hours a week overall. They worked on 79% of weekend days and 70% of vacation days. Meetings took 72% of their work time, leaving 28% alone. Strategy got 21%. They slept about 6.9 hours a night.

Notice where the extra hours came from: weekends and holidays. That is the rest day Pencavel's munitions workers lost.

Canadian owners are not far off. In Statistics Canada's 2023 Labour Force Survey data, 26.0% of self-employed people worked 49 or more hours a week, compared with 5.5% of paid employees. Self-employed people were 13.2% of everyone working but 41.5% of everyone working long hours. Among self-employed men, it was 31.5%.

And the strain is showing. BDC surveyed 1,500 Canadian small business owners in February and March 2023: 45% reported mental health challenges, up from 38% a year earlier, and work/life balance was a concern for 54%, up from 45%. BDC noted many owners were working even longer hours to stay afloat, though it did not measure how long.

If any of that sounds familiar, read the signs of founder burnout and the local picture in Vancouver founder burnout.

How many hours should a CEO work?

Here is my recommendation: plan for 50 hours a week, accept the odd 55 during a genuine crunch, and treat anything regularly above that as a problem to fix, not a badge.

Why 50 and not 40? Because I am not going to pretend founders run on 40, and the munitions data puts the bend in the curve around 48 to 49. Why not 60? Because the productivity gain past 50 is small, the stroke data starts climbing in the 49 to 54 band, and the 60-hour weeks Porter and Nohria found were built on working weekends and holidays.

You cannot manage a budget you have never measured, so start with the log.

Run a two-week honest hours log

For 14 days, write down every block of work in 30-minute chunks. Not what your calendar says; what you actually did. Count everything: email at 10 p.m., the Slack check on the sideline of your kid's game, the "quick" Saturday call.

Tag each block with one of five labels:

1. Decisions and strategy (thinking, writing, deciding)
2. Meetings with your team
3. External (customers, investors, partners, board)
4. Admin and email
5. Reactive (fires, interruptions, things someone else should own)

At the end of two weeks, total each category and the hours by day of week. The questions I would ask of that log are simple. How many hours were truly yours, alone and undistracted? How many days had zero work? Which reactive hours belong to someone else? Most people find the total is higher than they guessed and the decision hours are lower.

Build a 50-hour budget

Once you have the log, give the 50 hours a job before the week starts. A starting allocation I would suggest:

Adjust the split to your business. The structure matters more than the numbers: a hard total, a protected block of alone time, and a buffer with an edge. When a new commitment shows up, something else comes out of the budget. That is the whole discipline.

Protect one full day off

If you only take one thing from Pencavel's data, take this one. Losing the weekly rest day cost about 10% of output on its own, at the same weekly hours. The six-day, 48-hour week beat the seven-day, 70-hour week.

A day off means zero work. No "just checking email." Pick the day, tell your team, and set up coverage before it arrives:

• Name who decides what in your absence, in writing.
• Agree on what counts as a real emergency (and how they reach you, which should be a phone call, not a Slack message).
• Turn off work notifications on your phone for the day.

If a true emergency comes through more than once a quarter, that is a staffing or delegation problem, and it is worth fixing on its own. I wrote about why being permanently reachable fails as a leadership approach in always on is not a leadership strategy.

A shutdown ritual using the 5 still minutes

The 2% Solution, which I have taught since 2015, is 30 minutes a day: 15 moving, 5 still, 10 reading. The 5 still minutes are a natural end-of-day shutdown. Here is the routine I would recommend at the close of each workday:

1. Write down every open loop in your head, on paper or one note.
2. Pick tomorrow's first decision block and what goes in it.
3. Close the laptop and turn off work notifications.
4. Sit for 5 still minutes. No phone. Slow breathing, eyes closed or on a fixed point.
5. Say out loud that the workday is done, then leave the room.

It takes about ten minutes. The point is a clear line between work and the rest of your life, so the hours you did not budget do not leak back in at 10 p.m.

How to talk to your board or co-founder about it

Boards and co-founders worry that "working less" means caring less. So do not frame it that way. Frame it as risk management and performance.

With a board, I would bring three things. First, the two-week log, showing where the hours actually went. Second, the research in one page: the productivity curve flattening near 50 hours and the stroke risk linked to 55-plus. Third, a concrete plan: the 50-hour budget, the protected day off, who covers decisions that day, and a 90-day trial. Agree up front on what you will measure, such as decision turnaround time, progress on the two or three priorities the board cares about, and how many issues escalate to you. If those hold or improve, the trial becomes the policy.

It also helps to name the key-person risk plainly. A CEO with no recovery and no backup for decisions is a single point of failure, and boards understand single points of failure.

With a co-founder, make it mutual. Trade coverage days so each of you gets a full day off, share both logs, and agree on the emergency rule together. If one of you is carrying 65 hours while the other carries 45, the log will show it, and that is a conversation about roles, not stamina.

What to do this week

• Start the two-week hours log today, in 30-minute blocks.
• Pick your day off for next week and name your cover in writing.
• Block 12 hours of alone time on next week's calendar before anyone else books it.
• Add the 5 still minutes to the end of each workday.

If you want help building the budget around your actual week, the executive health page explains how I coach founders and executives one to one, or book a free 20-minute call and bring your log.