I recently came across John Pencavel’s summary of the British government’s efforts to understand worker productivity in munitions plants during World War I, and was surprised it isn’t more widely cited in articles about time management, productivity, and working hours. Certainly, I would have cited it in at least two things I’ve written if I’d known about it earlier. I decided to write a short post about it to help spread the word.
It seems a bit ridiculous to summarize a summary, though. The summary is well-written and quite accessible for the lay reader. I encourage you to go read it for yourself.
Instead of summarizing it, I will discuss it and how I think the findings it summarizes apply to the far different world of modern day “knowledge work.”
The data discussed in the paper dates from World War I. The British government found itself faced with ammunition shortages that could have an impact on the outcome of the war, and was therefore extremely motivated to determine how to achieve the greatest productivity from its munitions plants. The British Health of Munition Workers Committee was formed and charged with providing advice about the health and efficiency of munition workers. As part of this remit, the committee comissioned studies on the link between work hours and work performance.
The studies looked at several different factors, including overall hours and the difference between 7-day and 6-day work weeks.
For work hours, they found that the relationship between work hours and output is non-linear. Up until about 48 hours per week, output increases roughly in proportion to the number of hours worked. Starting at 49 hours/week, output rises at a slower rate. The maximum output is achieved at about 63 hours/week, and output in a 70 hour work week is almost the same as that in a 56 hour work week.
The data also clearly supported providing workers with a day of rest. Percavel’s analysis indicates that “[t]he loss in output from denying workers a day of rest is about ten percent.”
I was struck by how well the results in this old data set matched my own impressions of my own work capabilities. I once worked at a large contracting firm. We charged hours, meaning that we logged how we spent every 15 minutes in the day. I was paid a salary, though, so usually there was no financial incentive for me to work more than a standard 40 hour week. Sometimes I worked a bit more to keep my work on schedule, but I never tried to maximize my hours. This changed when I was assigned to a project that was already behind schedule. That project had been granted what were called “extended work week” hours. This meant that if I charged more than 40 hours per week, I would be paid for the extra hours. This was before I had children or other non-negotiable demands on my time outside of work. I was quite incentivized by the offer of extra money, and tried to maximize my hours.
However, I was also bound by our rules of ethics, which I, like everyone on the team, took very seriously. I discovered that even though there was plenty of work to do, I could not ethically charge more than about 55 hours per week. It just wouldn’t have been ethical to charge for hours in which I had the necessary work loaded on my computer, ready for my attention, but was unable to focus on the work. Essentially, I could not do more real work after roughly 55 hours in a given week, no matter how many hours I sat in front of my computer.
Over the years, I’ve paid more attention to the relationship between my productivity and my work hours, and have done several time-tracking exercises to better understand my work patterns. I consistently see a decline in marginal value of hours after about 45 work hours in a week. I can work more, and I even produce useful output, but I’m not producing it at the same rate I am in those first 40-45 hours. Worse, I notice more errors creeping into my work. In general, I don’t see much benefit from working more than 45 hours in a week, although I can do it for short periods of time if needed. Instead, I try to plan my work so that I can meet all of my deadlines without needing to push myself into that less productive zone.
Of course, that is all just anecdote. It is relevant to me, and I certainly use this knowledge when planning my time use. But it is not necessarily applicable to anyone else. To make general conclusions, we need some actual data. Sadly, we don’t really have it outside of studies like the one described above, which focus on more physical labor.
One obvious problem with attempting a study of knowledge worker productivity is that it is much more difficult to measure output. What is a unit of output for most of us? For me, it can be an article or blog post, a database design, a project plan, a set of training slides… or “just” an observation or realization that allows me to understand something more clearly. None of these things are uniform in size, nor do they have easily quantifiable quality metrics that would allow sub-par work to be neatly handled in a study. To be honest, I suspect asking me to rate my productivity on a scale of 1 to 10 would be more accurate than trying to count output.
That leaves us in a frustrating situation. It would be very useful to be able to say something general and meaningful on the relationship between hours worked and productivity, but our data are lacking. In addition to the WWI studies discussed above, we have additional old data from the early industrial age, and some more recent data looking at the impact of long hours on specific tasks relevant to many jobs, such as the ability to make sound judgement calls (summarized in the Harvard Business Review by Sarah Green Carmichael). We will probably never get a data set as clear cut and convincing as the British munition workers data set.
If we think about the data we do have and our own observations of our work patterns, though, I think the answer is clear. We know that mental fatigue is a real thing—afterall, we frequently complain about how tired we are at the end of a long day of… sitting at our computer. That is not physical exhaustion. If we’re honest with ourselves, we also know that we do not produce our best work when we’re suffering from mental fatigue, and that longer hours are not always productive hours. We should also know (because we actually have data on it) that even if our extra work hours are producing meaningful output, they come at a cost to our health and well-being.
Because of all of this, one of my goals when I manage a project is to run it in such a way that everyone can work healthy hours. I do this for the good of the project and for the good of the people on the team.
I wish more managers considered work hours when planning their projects. I also wish more people would put in the effort required to understand their own optimal work patterns. I think many people would be surprised to find that they’ve been working in a way that harms their health for no real gain in productivity. Yes, that can be a painful realization, particularly when there are years of unnecessary overwork to reflect upon. However, think about how the leaders of the British war effort must have felt when they realized that their attempts to increase productivity at munitions factories were backfiring. If they could face what the data told them and change course, so can we.