Sleep and Human Capital: Evidence from Daylight Saving Time

This paper is one of the first to test for a causal relationship between sleep and human capital. It exploits the quasi-experimental nature of a simple policy regulation that has been affecting the sleep pattern of more than one billion people in 70 countries around the globe: “Daylight Saving Time (DST)” setting clocks forward one hour in spring and backward one hour in fall. First proposed by Benjamin Franklin to save candlelight in a satire letter to the Journal de Paris, Germany and Austria-Hungary were the first countries to introduce DST during World War I (WWI). Under DST, during the summer, people spend more time with natural daylight in the evening; this aligns with most people’s preferences. Today, all countries in the European Union, the great majority of the US states and Canadian provinces, as well as 40 other countries such as Mexico, Chile, Israel, and Iran set their clock one hour forward in spring and one hour back in fall.

The paper uses two datasets to study the human capital effects of mild changes in sleep pattern due to DST: (a) The US Behavioral Risk Factor Surveillance System (BRFSS), and (b) The German Hospital Census, which provide empirical evidence from the biggest American and the biggest European country. In total, the representative databases cover the behavioral reactions of 400 million individuals in Europe and the US. Each database covers the entire year—not just spring or fall DST clock changes—over the first decade of the new millennium. Both representative datasets carry a very large number of observations. This is crucial to control for important seasonal confounding factors while maintaining enough statistical power to identify health effects at the daily level. Because our estimates rely on both up to 3.4 million individual observations from the US (BRFSS), and up to 160 million observations from Germany (German Hospital Census), they are very precise. In addition, the two databases complement each other: While the BRFSS captures the entire population and contains a rich battery of self-reported health, satisfaction, and sleep measures that elicit mild(er) human capital effects of changes in sleep pattern, the Hospital Census captures severe health effects that require inpatient stays.

We find evidence of mild negative health effects when clocks are set forward one hour in spring. When clocks are set back one hour in fall, effectively extending sleep duration for the sleep deprived by one hour, sleep duration and self-reported health increase and hospital admissions decrease significantly for four days. In summary, this study is one of the first to show that even mild changes to sleep patterns can affect human capital in significant ways. We demonstrate that more sleep may lead to significant, immediate, health improvements for people on the margin to getting hospitalized.