Simulating JOLTS Hires and Separations Data Using Historical QCEW Data

Mark Crankshaw


The Bureau of Labor Statistics produces the monthly JOLTS survey which is a 16,000 unit sample of business establishments drawn from a population frame of over 8 million establishments. While the JOLTS sample is allocated and selected with the goal of accurately reflecting the general composition of the population, the JOLTS sample does not reflect the general composition of the population with respect to the establishment age. If these very young establishments systematically differ from relatively older establishments with respect to hires and separations rates, then the JOLTS hires and separations rates may be biased. A simulation of JOLTS data using historical QCEW employment data was developed to determine the nature and extent of the bias. It was found that JOLTS respondents systematically underreport separations, the JOLTS sample is over-representative of older more stable firms.