Are labor market indicators telling the truth? : role of measurement error in the U.S. current population survey

Labor market indicators are critical for policymakers, but measurement error in labor force survey data is known to be substantial. In this paper, I quantify the implications of classification errors in the U.S. Current Population Survey (CPS), in which respondents misreport their true labor force status. Once I correct for measurement error using a latent variable approach, the unemployment rate is on average 0.8 percentage points (ppts) higher than the official unemployment rate, with a maximum of 2.0 ppts higher between 1996 and 2018. This paper further quantifies the contributions to business-cycle fluctuations in the unemployment rate from job separation, job finding, and participation. Correcting for misclassification changes previous studies' results about the contributions of these transition probabilities: job separation accounts for more of the unemployment fluctuations, while participation accounts for fewer. The methodology I propose can be applied to any other labor force survey in which labor force status is observed for three periods