Trends in Hours Worked across Countries: Evidence using Micro Data

We use three different micro data sets, the European Labor Force Survey, the Current Population Survey, and the German Microcensus, to obtain annual hours worked for various demographic subgroups in the US and 18 European countries. One major difficulty in constructing annual hours from micro data sets is the fact that hours are only reported for selected reference weeks, which are not distributed evenly over the year. As a consequence, hours lost due to public holidays or annual leave are mis-measured. Consequently, we use external data sources to account for these hours. We compare the generated time series of aggregate annual hours worked to the data series provided by the OECD and the Conference Board. In 29 out of 38 cases, average deviations amount to less than 10 percent, in 20 out of 38 cases to less than 5 percent. It is worthwhile pointing out that there are also non-negligible differences between the OECD and the Conference Board. We also decompose hours worked in an intensive and an extensive margin, and compare both to OECD and Conference Board data, as well as checking internal consistency of different possible extensive margin definitions. Finally, we document a set of stylized facts on hours worked (conditional on working and unconditional) over time and across countries for different demographic subgroups distinguished by marital status, gender, and age.