We all know sleep is essential to ensure we have enough energy during the day. Yet a lot of people suffer from sleep deprivation. This can be a huge issue as it can lead to lower productivity and lack of alertness, which may even culminate in lower economic growth.
But forget about economic growth for the moment, let’s concentrate on something that individuals may care much more about: life satisfaction. Does sleep duration affect life satisfaction? You bet. In fact, it turns out the average individual sleeps about an hour less than what would maximize their life satisfaction.
This is one of the conclusions of a paper by Piper (2015). But let’s not get ahead of ourselves. Let us look at what the existing literature says.
First, the medical literature has identified that sleep has U-shaped correlations with various health-related variables. This means that those who sleep too little or too much tend to be less healthy. Adverse health effects could include weight gain, obesity, diabetes, hypertension, or even deteriorated mental health.
Second, there is an economics of sleep literature. This literature usually models sleep duration as a decision whose outcome depends on how one values the trade-off between high alertness/productivity and more time for leisure/work. I.e. sleeping more makes you more productive allowing you to perhaps have a higher income and thus consume more; but sleeping less allows you to spend more time on leisure activities (which is fun), or work more (which increases income and consumption again). A standard model in this literature would predict that high-income people would choose to sleep less.
Now, let’s move on to Piper’s (2015) study, the subject of this post. The author has a sample of around 24,000 German individuals who were followed up on for several years. So overall, the sample includes around 70,000 person-year observations. The question we’re after is whether sleep duration affects life satisfaction and if yes, how.
Individuals in the sample sleep on average 7 hours, which is within the bounds recommended by the National Sleep Foundation (7-9 hours for adults), but barely. Also, given that this is an average it’s quite likely that a substantial fraction of individuals sleep less than recommended.
Sleep duration varies interestingly with various demographic variables. Below are four charts showing average sleep duration on a weekday by various subsamples.
We can see that economic models of sleep are likely correct: high income people sleep less. Healthier people sleep more (an interesting clue), kids make one’s time disappear and thus decrease sleep duration, while age and sleep have a U-shaped relationship.
But what about sleep’s effect on life satisfaction? Not surprisingly, Piper finds a hump-shaped effect. More sleep is desirable until a certain point, from which it negatively affects life satisfaction. The big question is what that critical point is?
The amount of sleep associated with maximum life satisfaction (and thus the critical point) is estimated to be somewhere between 7.9 and 8.3 hours a night. The former comes from a regression with individual fixed effects, the latter from a pooled OLS specification. In any case, this implies that the average person in the sample sleeps about an hour or so less than what would maximize their life satisfaction.
There is some heterogeneity in subsamples but not a lot. The most important is that females need on average only 7.4-8.1 hours of sleep to maximize life satisfaction as opposed to 8.6-8.7 hours for males.
These are intriguing results, especially given that the sample average sleep duration is only 7 hours. But just how much life satisfaction could one “gain” by switching to the prescribed roughly 8 hours of sleep instead of e.g. 7? On a scale of 0-10, one’s life satisfaction would increase by a mere 0.06 points (or about 1/5 the difference between married and single people) if one were to sleep 8 hours instead of 7. This is rather small, but going from 6 hours to 8 hours would increase life satisfaction by 0.26 points, which is roughly equivalent to how much happier married people are than singles. So the effects are not negligible.
The figure below plots how many life satisfaction points one is expected to gain if they changed their sleep duration from various hours (“initial hours slept”, measured on the x-axis) to 8 hours. We can see that the two models estimated by the authors give somewhat different results, the truth likely lies somewhere in-between the two curves. For reference, I also plotted the “marriage premium” implied by both models, that is by how many points married people are happier than single people.
So, should we all now change how much we sleep? Probably no. The amount of sleep needed varies from person to person, not the least because several genes have been identified to contribute to it in one way or another. So there probably is no one-size-fits-all solution.
But what’s concerning about this study is that the average of a large representative German sample is so much below the optimal amount of sleep. And that it is at the lower bound of the recommended range. As mentioned above, this quite likely implies that a substantial fraction of the population sleeps less than the average (less than 7 hours), and these people could potentially gain by changing their sleeping habits. And quite likely, with higher life satisfaction, they might be more productive, and lots of small gains could add up and even boost the economy as a whole.
This study also shows that perhaps on average, people undervalue sleep in today’s society, and they lose life satisfaction because of this. So perhaps, those sleeping less than 8 hours a night could give a little more sleep a try.