Comparing Hours per Job in the CPS and the ATUS

Abstract

The number of hours that people work is a key economic and social indicator. Thus, it is important to know whether household surveys measure hours worked correctly. In earlier studies (Frazis and Stewart 2004, 2007, 2008), we examined how hours worked in the Federal government's main economic household survey, the Current Population Survey (CPS), compare with time-diary data from the American Time Use Survey (ATUS). Most of our previous work concentrated on comparing hours worked per person per week for employed persons (Frazis and Stewart 2004, 2007). In Frazis and Stewart (2008), we compared hours worked per job for a particular class of jobs in an attempt to replicate the Bureau of Labor Statistics (BLS) establishment survey, the Current Employment Statistics (CES) survey. In this article, we compare weekly hours worked per job for all jobs in the ATUS and the CPS. Hours worked per job are of interest for a variety of reasons. Studies of hourly wages are usually restricted to main jobs because many surveys do not collect earnings data for second jobs. Official productivity measures mainly use data from the CES (which are per-job estimates), but for jobs not covered by the CES?supervisors, nonproduction workers, and the self employed?productivity measures use data from the CPS.