A Landmark Study Found the Exact Point Where More Hours Stop Helping
Stanford research spanning a century of labor data found output rises with hours worked only up to a point, then flattens, then drops. Seventy hours a week produced no more than fifty five did.
More hours should mean more output. It is one of the most intuitive assumptions in how work gets structured, and it is also, according to one of the more thorough bodies of labor economics research available, wrong past a fairly specific and fairly low threshold.
Stanford economist John Pencavel spent years studying the actual relationship between hours worked and output produced, using detailed historical records, most notably British munitions workers during the First World War, whose output could be measured with unusual precision since factories tracked individual production closely for wartime planning. Published in The Economic Journal in 2015, his analysis found the relationship is sharply nonlinear. Below roughly 49 to 50 hours a week, output rises close to proportionally with hours worked, additional time genuinely buys additional output. Past that threshold, output continues to rise, but at a steadily decreasing rate, and past roughly 55 hours, the curve flattens so completely that a worker putting in 70 hours a week produced, on average, no more total output than one working 55 to 56 hours. All of those additional 15 hours purchased essentially nothing.
Pencavel described this as a highly nonlinear effect, noting explicitly that an extra five hours added to a 35-hour week behaves completely differently than an extra five hours added onto a 48-hour week, a distinction that gets lost in simple advice to just work more when a deadline is looming. His broader research, later expanded into a full book on the subject titled Diminishing Returns at Work, also found the costs of long hours extend beyond the point of zero marginal output. Extended hours were associated with a measurably higher rate of workplace accidents and health problems, in both blue-collar and white-collar populations, meaning the hours past the productivity ceiling were not merely wasted. They were actively adding a real, separate cost on top of producing nothing extra in return.
It is worth being fair to the limits of century-old munitions data applied to modern knowledge work, and later researchers studying more contemporary settings, including call centers and other white-collar contexts, have generally found supportive though not identical patterns, with some estimates placing the productive ceiling for knowledge work somewhat lower, closer to 38 to 48 hours a week depending on the specific job. One study using changes in mandatory overtime laws for nurses even found a case running the other direction, where added overtime hours reduced care quality through downstream staffing effects rather than through the same mechanism Pencavel identified, a reminder that the precise number moves around by context and by industry. The underlying shape of the curve, a real ceiling beyond which more hours stop paying for themselves, holds up consistently across the studies that have tested it.
This offers a genuinely useful, evidence based counter to the instinct that a demanding stretch always calls for simply working longer. Occasional short bursts past the ceiling are not catastrophic and sometimes genuinely necessary. Treating extended overtime as a sustainable, default strategy runs directly against a substantial, century-spanning body of research showing that past a fairly identifiable point, it stops functioning as an investment in output and starts functioning as a cost with no return attached to it at all.