Why job openings can outnumber hires

Employers had 7.3 million job openings in July 2026, while recording 5.1 million hires and 3.1 million quits. The Bureau of Labor Statistics described each as little changed. Published September 1, these Job Openings and Labor Turnover Survey (JOLTS) figures cover the US nonfarm sector. They describe July, not August hiring.1

The apparent mismatch starts with the calendar. Openings are a snapshot on July’s last business day; hires count payroll additions throughout July. JOLTS asks establishments—reporting workplaces—for separate totals, without following individual vacancies into hires. Consequently, subtracting the totals cannot measure rejected applicants or a recruitment backlog. The month-end snapshot is not the starting inventory for that month’s hiring.32

An opening must meet three conditions: a specific position and available work exist; the job could start within 30 days, whether or not a suitable candidate is found; and the workplace is actively recruiting workers from outside that location. That 30-day condition describes when work could begin, not a promise to hire by then. Internal-only openings, including promotions and transfers, are excluded, as are positions starting more than 30 days ahead. Positions already filled by employees who have not yet reported for work are excluded, too. A position awaiting someone’s arrival therefore does not automatically qualify as an opening.2

Hires count all payroll additions before departures are deducted—the meaning of gross hiring. They include rehires and workers both hired and separated within the month. Those inclusions matter: a payroll entry need not remain through month-end. Hires therefore do not measure newly created jobs, nor does this total quantify replacement hiring.2

Subtracting total separations, meaning payroll departures, from hires produces JOLTS implied employment change. The separate Current Employment Statistics survey measures payroll employment change. BLS says the measures track well over time, but different definitions, reference periods and estimation methods can produce short-term divergence. One cannot simply substitute for the other.32

Quits are voluntary payroll departures, excluding retirements and transfers to other locations. BLS says the quits rate can indicate workers’ willingness or ability to leave. That qualified interpretation matters: voluntary departures do not establish that workers found better jobs or higher pay.21

The national total also leaves room for different industry experiences. Professional and business services recorded 188,000 fewer hires than in June despite little-changed hiring overall. That identifies a sector’s movement beneath the aggregate; it does not explain employers’ reasons or establish what happened to particular occupations or workers.1

The percentages preserve these distinctions. The openings rate expresses openings as a share of employment plus openings: filled and unfilled jobs together. Hires and quits rates express monthly activity relative to employment. Employment refers to the pay period including the 12th, a different reference point from month-end openings.13

These totals are survey estimates, and some answers are incomplete. When an establishment reports only part of the requested information, BLS estimates missing items from available data, a process called imputation. It groups establishments by whether employment is expanding, stable or contracting, then uses reported information within those groups. For an expanding workplace, the estimate draws on information from other expanding workplaces; stable and contracting workplaces receive corresponding treatment. This connects the missing-item estimate to similar employment dynamics. It supplies an estimate informed by respondents, while the missing answer remains something the survey did not directly observe.4

Precision has other limits. Sampling error arises because surveying a sample can yield a different result from surveying the whole population. Nonsampling error includes missing or late responses, reporting mistakes and errors in collecting or processing data. These are distinct sources of uncertainty: even correctly recorded responses from a sample need not reproduce the population’s exact total, while mistakes can affect the responses themselves. BLS evaluates differences using standard errors, statistical measures that help assess whether a change exceeds the variation expected from sampling. A numerical movement alone is therefore insufficient to establish a statistically meaningful change. The release’s description of little change should not be read as literal equality.51

July’s first estimates are preliminary. BLS revises them the following month to incorporate additional responses. Later information can change the reported figures.3

Report an error · Corrections policy