You work forty hours a week, near enough, and the number has the quiet authority of a fact of nature. It is the size of a full-time job, the thing a salary is quoted against, the line past which the week turns into overtime. Ask why it is forty and not thirty or fifty and the question can sound almost like asking why a day has twenty-four hours. But a day is set by the turning of the earth, and the working week is not set by anything of the kind. It was argued over, legislated, and revised; it was a different length within living memory, and it is a different length right now in countries no less modern than this one. This chapter is about where the forty-hour week came from, why it rests where it does, and whether anything in economics fixes it there. The other half of the question a person tends to ask about their hours, whether they would come apart without them, whether the week is holding something together that nothing else would, is taken up later in this volume, in the chapter on work and purpose.
The long fall
Start with the long view, because the forty-hour week is the settled-down end of a steep and lengthy decline. In the rich countries of the late nineteenth century the working year ran to roughly three thousand hours. A German or American factory hand in 1870 put in something close to sixty hours across a week, in a year with almost no holidays. Over the century that followed it fell by nearly half.
The fall was not smooth, and the two halves of the picture are not measured the same way, which the figure marks with a break rather than a join. The early series counts the hours of full-time production workers, the people in factories and workshops for whom a week meant a fixed run of long days. The later series counts something broader: total hours worked divided by everyone holding a job, which now takes in part-timers, the salaried, and a workforce far more mixed than the one the early figure described. The two cannot be strung into a single line, and the war years between them left no usable numbers at all. What survives the break is the trend. Counted either way, the working year got shorter, generation after generation, across every rich country for which the record exists.
Behind the fall sits the rise in output per hour. As each hour of work came to produce more, part of the gain was taken as more goods and part as more time, and for a long stretch the second part was large: the shorter week was one of the things a growing economy bought with its growing productivity. The question of who captured the rest of those gains, and why the typical wage did not simply rise to match the output the typical hour produced, is the subject of the earlier chapter on why paychecks do not track the growth headline. The point that matters here is narrower: for roughly a hundred years, a real share of rising productivity was drawn down as time off the job rather than money in the packet, and that is what pulled the working year down toward the level it now sits at.
The fork
If shorter hours were simply what a rich country does, the lines would have converged as countries grew rich together. They did not. Germany and the United States began the period within a few hundred hours of each other, both near three thousand a year, and both fell; but by 2023 the German in work averaged about thirteen hundred hours across the year while the American averaged nearly eighteen hundred, a gap of a full working month and more. Korea, which industrialized later and faster, worked longer than anyone at the height of its boom, past two thousand nine hundred hours a year around 1970, and even after a steep decline still averages more than either. The spread is wide, and it did not narrow. Countries at a similar level of wealth run their working years hundreds of hours apart, which is the first sign that the length of the week answers to something other than a common necessity.
The fork runs inside countries as well as between them. The forty-hour week describes the middle of a labor market more than its edges. Salaried professionals routinely work well past it without the extra hours ever showing up as overtime, while someone holding two part-time jobs can exceed it without any single employer ever recording a long week. Who actually gets to work close to forty hours, rather than well above or well below, tends to track the same thing that set their pay: how easily they can be replaced and how credibly they can decline the extra hours. That is this chapter’s reading of the fork rather than a measured result here; the volume’s earlier chapters traced the bargaining machinery behind it. The forty-hour figure is a real average, but it is an average laid over a range, and the range is patterned by bargaining position rather than scattered at random.
Long hours do not buy proportional output
There is a second reason to doubt that the number is fixed by necessity, and it shows up when hours are set against what an hour produces. Across the rich economies, the countries that work the longest hours are not the ones that produce the most in each of them. The relationship runs the other way.
The countries that put in the longest years are not the productive powerhouses; they are more often the ones catching up. Germany, near the bottom of the hours range, produces more than twice as much per hour as Mexico, near the top of it. A snapshot like this cannot say which way the causation runs, and it does not need to for the point at hand. If long hours were the price of a rich country’s output, the longest-houred countries would be the richest per hour, and they are the opposite. The length of the working year cannot be read off from what an economy needs to produce, because economies that produce a great deal per hour choose, on this evidence, to work fewer of them.
Where the number comes from
So where does forty come from? Not from a calculation. The specific number is the residue of a long political fight and the laws it eventually produced. The demand for the eight-hour day, under the old slogan of eight hours for work, eight for rest, and eight for what you will, was a rallying cry of the labor movement across the nineteenth century, and eight hours across a week that shrank from six days to five is most of the arithmetic that lands on forty. In the United States the forty-hour week was written into federal law by the Fair Labor Standards Act of 1938, which allowed longer hours while requiring extra pay beyond forty, making forty the point at which an hour starts to cost an employer more. Other countries drew the line elsewhere and by other means. France legislated a thirty-five-hour standard week. Much of the German difference has been negotiated time off rather than a shorter working day: by Huberman and Minns’s accounting, the extra vacation days a German worker had won corresponded, by 2000, to almost half of the gap in annual worktimes between Germany and the United States. The line sits where each country’s history of law and bargaining left it, which is why it sits in a different place in each.
Does economics fix the number where the law and the unions left it? The theory says less than the confidence of the phrase “the forty-hour week” suggests. Standard labor economics treats hours as the outcome of a trade between income and time: a worker gives up leisure for pay up to the point where one more hour of pay is worth less than the hour of life it costs, and where that point falls depends on the wage, on tastes, on what the pay can buy, and on how much say the worker has over the trade at all. None of that pins a particular figure, and it does not pin the same figure for two different people or two different countries. That economics does not fix the length of the week is the reading this chapter draws from the spread and the history taken together, rather than a separate result of its own; but the spread and the history are hard to square with any claim that a single number is written into the nature of work. What the theory offers is a set of forces that push hours around, and those forces have pushed them to different places.
The 40-hour week is the natural shape of work.
Oversimplified Moderate confidence
The claim has a real kernel, which is why the ruling is not that the opposite is true. The length of the working week is not arbitrary. It responds to real forces: as output per hour rose, hours fell, because part of a richer economy’s gain is taken as time; and there are limits to how long people can usefully work, so no rich country runs a hundred-hour week or a ten-hour one. Hours are shaped, not scattered at random. What the claim then does is freeze that shaping into a single natural constant fixed at forty, and that step does not survive the evidence. The working week ran close to sixty hours when the record opens in 1870, and the working year has since fallen by nearly half. It stands today at the equivalent of a shorter week in Germany and a longer one in Korea and Mexico, with no sign of settling to a common figure. And it was set, where it is set, by an eight-hour movement, a wage floor on overtime, and national laws that drew the line in different places, rather than by any calculation of what work naturally requires. The evidence that long hours fail to buy proportional output cuts the same way: if forty were a natural optimum, the economies working near it would not be so easily out-produced per hour by the ones working well below it. So the claim is not backwards, because the week is shaped by real forces and is not arbitrary; it is oversimplified, because it turns a negotiated settlement that has moved before and differs across borders into a law of nature fixed at one figure. What holds the confidence at moderate rather than high is that “the natural shape of work” is a loose and partly normative idea rather than a measured quantity, so the evidence bears on it strongly without being able to settle it to the digit.
Sources
- The fall: hours per worker in the rich countries fell from roughly three thousand a year in 1870 to between about thirteen hundred and nineteen hundred today. Historical figures (full-time production workers, 1870–1938) from Huberman and Minns, Explorations in Economic History 44(4), 2007, Table 3; modern figures (hours per person employed, 1950–2023) from Penn World Table 11.0 — the two are different constructs and are not joined into one series. Germany 2023 about 1,335 hours, the United States about 1,789, Korea about 1,910.
- The fork and the output test: the countries with the longest hours are not the ones with the highest output per hour. OECD, average annual hours worked and GDP per hour worked (current PPP), 2024: Germany about 1,334 hours at $101 an hour of output against Mexico’s about 2,215 hours at about $40. The OECD warns its hours series are built for tracking trends over time rather than ranking levels across countries in one year; the gap between the extremes is far larger than the differences that caveat describes, and the comparison is read as descriptive, not causal.
- The number itself: the forty-hour week is a legal and bargained settlement. In the United States, the Fair Labor Standards Act of 1938 set the 40-hour threshold for overtime pay; France legislated a 35-hour standard week; the extra German vacation corresponded, by 2000, to almost half of the Germany–US gap in annual worktimes (Huberman and Minns). Standard labor-supply theory treats hours as an income-versus-time trade whose outcome depends on wages, tastes, and bargaining power, and fixes no particular number. The claim that economics does not pin the figure is this chapter’s own reading of the spread and the history together, offered as an inference rather than a separate finding.
- Confidence is moderate under the rubric, which scores the weaker of evidence directness and construct match. Evidence directness is strong: the hours are measured, the fall and the fork are read straight from the series, and the settlement is a matter of statute. The binding, weaker leg is construct match. “The natural shape of work” is a loose and partly normative notion rather than a measured quantity, so the data can weigh against it heavily without disproving it to the digit. The direction is not in doubt, which is why the ruling is Oversimplified and the confidence is not low; the exact reach of the word “natural” is, which is why it is not high.
A settlement, not a constant
The forty-hour week, then, is a settlement and not a constant. It is where a century of rising productivity, a long fight over the working day, and a particular set of national laws happened to come to rest, and it rests in a different place in France or Germany or Korea. That it feels natural is a measure of how settled it is, not of how fixed. Settlements can move, and they have moved before, which is worth holding onto against the sense that the shape of the working week is simply given.
The next chapter takes up a force that has moved the working week before and may again, and that carries its own long history of alarm. Machines have been taking over pieces of human work for two centuries, and each wave has arrived with the fear that this time there would not be enough work left to go around. The chapter asks whether the automation panics were wrong about jobs, and whether artificial intelligence is different.