|| Checking for direct PDF access through Ovid
From 2003 to 2012, time allotted for sleep in the United States is countercyclical with respect to employment conditions.Each additional point of state unemployment is associated with 0.83 more minutes of sleep, adjusted for secular trend.Higher state unemployment is associated with less frequent reports of sleeplessness, adjusted for secular trend.Recent research has reported that habitually short sleep duration is a risk factor for declining health, including increased risk of obesity, diabetes and coronary heart disease. In this study we investigate whether macroeconomic conditions are associated with variation in mean sleep time in the United States, and if so, whether the effect is procyclical or countercyclical. We merge state unemployment rates from 2003 through 2012 with the American Time Use Survey, a nationally representative sample of adults with 24 h time diaries. We find that higher aggregate unemployment is associated with longer mean sleep duration, with each additional point of state unemployment associated with an additional average 0.83 min of sleep (p < 0.001), after adjusting for a secular trend of increasing sleep over the time period. Despite a national poll in 2009 that found one-third of Americans reporting losing sleep over the economy, we do not find that higher state unemployment is associated with more sleeplessness. Instead, we find that higher state unemployment is associated with less frequent time use described as “sleeplessness” (marginal effect = 0.05 at 4% unemployment and 0.034 at 14% unemployment, p < 0.001), after controlling for a secular trend.