We already knew that, in most professions, people work to the point of negative productivity – in the sense that had they worked less they would have produced more. Myself and a coauthor even have a paper about this, in which we speculate that workers signal their ability at the expenses of current output - think about signing up for a new project when you are already fully committed to other stuff and unable to properly do it, in order to add an extra line to your CV and become a better candidate for a future promotion or future job.
Unsurprisingly, I very much enjoyed a recent article which nicely summarizes the scientific evidence regarding the optimal length of the work week (as it turns out, it is around 40 hours). Thanks to the article, I also discovered some interesting studies dealing with the perception of productivity. As it turns out, people are quite bad at measuring objectively the quality/quantity of their output, and they become worse at it the more they work. Hence, as they extend their work week, they think they are still producing something valuable well beyond the point in which they stop being productive.
This evidence seems to confirm casual observation. However, the puzzle remains: workers' judgment is bad in many dimensions (estimating risks, staying awake, …) and there are many systems to compensate for this. Think about systems to prevent drivers from dozing off, or mandatory health and safety regulations. So the question is: why are long working hours actually encouraged in many workplaces, if not explicitly via peer pressure? Isn't the problem rather that employers are bad at estimating their workers' productivity?
UPDATE: an interesting development from Sweden.