Reassessing classification errors in the analysis of labor market dynamics

Classification error–misreporting of true labor force statuses by survey respondents–can impact our broader understanding of labor markets in two critical dimensions: (i) measurement of labor market stocks (e.g., the unemployment rate); and (ii) the flows between these states (e.g., sources of unemployment fluctuations). This paper makes three contributions to the literature. First, this paper provides a single unifying framework for the classification error of labor market states, which can be used to analyze both stocks and flows. Second, I compare the proposed model implications with other classification-error correction models and show that the proposed latent variable approach generates similar labor market dynamics as re-interview-survey-based methods. Third, I show that correcting for classification errors with this alternative approach supports but moderates the results of previous studies in terms of the measurement of the unemployment rate and sources of unemployment fluctuations. It (i) mutes the magnitude of an underestimate of the U.S. official unemployment rate and (ii) points to a more (less) prominent role of the job separation margin (the labor force participation margin) in explaining the unemployment fluctuations.