The Overworked American or the Overestimated Workweek? Trend and Bias in Recent Estimates of Weekly Work Hours in the United States

Abstract

Individuals' estimates of their paid working hours in response to the Current Population Survey (CPS) are biased upward. The bias is demonstrated here using two independent alternative sources of work-hour estimates: time-diary studies and the BLS Current Employment Statistics (CES) survey, based on payroll records. Comparison of the CPS and CES suggests that the individual overestimation of hours increased substantially between 1965 and 1990, an increase that cannot be attributed to obvious differences between the CPS and CES in coverage or treatment of moonlighting. Examination of the distribution of reported hours in the CPS shows that the increase in bias is not a consequence of the heaping of individual hours responses on the 40-hour week. Crosssection analysis of 1985 time-diary data indicates that hours overestimation varies systematically by gender, education level, and presence of young children. The changing composition of the workforce by schooling can explain about half the increase in the overestimation of hours.