Job Openings and Labor Turnover Survey
The Job Openings and Labor Turnover Survey (JOLTS) is a continual survey operated by the BLS.
Design
Design
A sampling frame of firms is collected from administrative data. The frame is then stratified by ownership (private or public), Census region, industry, and size, creating 37 panels. 1 of these is a 'certainty' panel; private firms with 5,000+ employees are sampled with certainty. Otherwise, sample is Neyman allocated.
A fresh panel is selected every year. Selected firms are empaneled for 3 years (used to be 2?), after which they cannot be re-selected again for 3 years. The certainties are the exception to this rule.
Simultaneous to selecting a fresh panel, the old panels still in collection are re-stratified. Firms that have closed are dropped. Finally, sample weights are re-calculated and post-stratified.
Every quarter, 'birth samples' are selected among firms reporting their first employees. The set of such firms are stratified by age, industry, and size. The sample size allocated to each stratum is calculated as the number of 'births' divided by the stratum re-weight. The sample weight for selected units is simply the inverse selection probability.
Administration
A small portion of selected firms are dropped from the panel during 'address refinement'. Then approximate 10% of the panel is lost at enrollment.