JOLTS Explained: Job Openings, Hires, and Quits

The U.S. Bureau of Labor Statistics (BLS) surveys establishments about job openings, hires, and separations. The purpose is to describe labor demand and movement into and out of payrolls. JOLTS reports levels and rates for the total nonfarm sector, industries, establishment-size groups, and selected geographic areas. The [JOLTS definitions]({source:blsJoltsDefinitions}) and [frequently asked questions]({source:blsJoltsFaq}) explain the survey's concepts.

These are employer-based statistics. They do not count unemployed people, job seekers, or advertised listings one by one. The monthly [JOLTS release]({source:blsJoltsCurrentRelease}) reports estimates from a sample, which means readers should pay attention to reference periods, seasonal adjustment, and revisions.

JOLTS counts a job opening only when all three conditions hold on the last business day of the reference month: a specific position exists and work is available; the position could start within 30 days; and the employer is actively recruiting from outside that establishment. A plan to hire later, an internal promotion opportunity, or a position already filled does not meet the definition. The [BLS opening criteria]({source:blsJoltsFaq}) spell out what active recruiting means.

The level is a stock: it measures qualifying unfilled positions at one point in time. It is not the number of job advertisements seen during the month, and one employer may have more than one opening. Openings can go up because employers create or post positions, or down because they fill, cancel, or otherwise remove them from the qualifying set. The count alone does not say which event caused the change.

JOLTS hires are additions to an establishment's payroll during the entire reference month. The category includes new and rehired employees, full- and part-time workers, some recalls, hires who separate again during the same month, and transfers from another location of the same firm. It excludes transfers or promotions within the reporting location and workers supplied by outside temporary-help or leasing firms. See the [BLS definition of hires]({source:blsJoltsDefinitions}).

Hires are a flow: they count events across a period, not positions sitting open at one moment. A hire reported this month need not correspond to a job opening still present at month-end. A position might have been opened and filled earlier, or the addition might be a recall or inter-location transfer. Conversely, an opening on the final business day may not be filled until a later month.

Quits are separations initiated voluntarily by employees during the month. They exclude retirements and transfers to other locations of the same firm, which JOLTS records as other separations. Total separations combine quits, layoffs and discharges, and other separations. Layoffs and discharges are involuntary events initiated by the employer; other separations include retirements, deaths, disability, and qualifying transfers. The [JOLTS separation definitions]({source:blsJoltsDefinitions}) distinguish these categories.

A higher quits rate can be consistent with more workers willing or able to leave jobs, but it is not a direct measure of worker satisfaction or a count of people who successfully moved to better jobs. Quits can reflect opportunities, pay, conditions, personal circumstances, and other factors. Read the level, rate, industry pattern, and broader labor-market context together.

The openings level is measured on the last business day; hires and separations cover the whole month. That stock-versus-flow difference alone means they are not expected to match. Their definitions also differ: hires include some recalls and transfers from other locations, while openings require active external recruitment for a position that could start within 30 days.

BLS notes that hires minus separations gives an implied employment change that is conceptually similar to, and tends to track, the monthly change in Current Employment Statistics (CES) employment. The two can still diverge in the short run because their definitions, reference periods, and estimation procedures differ. Do not treat JOLTS hires minus separations as an exact payroll-jobs reconciliation. The [JOLTS FAQ]({source:blsJoltsFaq}) discusses this comparison.

The aggregate series also do not match each hire to a particular opening. An opening is a qualifying vacancy still present on the last business day; a hire is a payroll addition recorded at any point in the month. Even if the two counts move together, the difference does not reveal how many month-end vacancies were filled, how long they stayed open, or which worker took which job. Those are questions about linked worker and vacancy histories, not a subtraction of two published JOLTS totals.

<!-- learn:illustration -->

A level is a count of openings or payroll events. A rate relates a count to an employment base, making comparisons across establishment groups of different sizes more meaningful. JOLTS uses different denominators for openings and the other event rates:

Job openings rate = job openings ÷ (employment + job openings) × 100

Hires, quits, and other event rates = event count ÷ employment × 100

The [BLS rate definitions]({source:blsJoltsFieldDefinitions}) specify these denominators. Because the job-openings rate includes openings in its denominator and the hires and quits rates do not, those percentages are not calculated from the same base. A reader comparing a 4.5% openings rate with a 3.5% hires rate should not conclude that exactly one percentage point of positions remained unfilled.

Suppose a fictional establishment group reports 100,000 employees, 5,000 job openings at month-end, 4,000 hires during the month, and 2,000 quits during the month. These amounts are illustrative, not BLS estimates.

The different openings-rate denominator explains why the three rates cannot be compared as though they were shares of one identical pool. The example also does not imply that 4,000 hires filled 4,000 of the 5,000 month-end openings. The hires occurred throughout the month, while openings were counted at its end.

Rates can also move because their employment denominator changes. In the same hypothetical group, 2,000 quits among 100,000 employees is 2.00%. If quits stayed at 2,000 while employment rose to 105,000, the rate would be about 1.90%. The lower rate in this example would not mean that fewer people quit; the numerator is unchanged. When a JOLTS rate moves, check both the event count and its denominator.

First identify the measure and reference period. Openings refer to the final business day; hires and separations refer to the full month. Then check whether the table is seasonally adjusted or not seasonally adjusted and compare it with a series on the same basis. Levels tell you about counts; rates account for differences in the employment base.

The first monthly estimate is preliminary. BLS revises it the following month as more establishment reports arrive and seasonal factors are recalculated. Annual updates can revise several prior years. A sharp one-month movement deserves a check against the revision history and related data before it becomes a broad claim about labor-market strength. The [JOLTS FAQ]({source:blsJoltsFaq}) describes its revision process.

JOLTS is useful alongside, not in place of, other labor indicators. CPS measures people's employment status and supplies the national unemployment rate; CES estimates payroll jobs and earnings. UI claims track administrative applications for benefits. The [CPS concepts]({source:blsCpsConcepts}), Beveridge curve guide, CES versus CPS guide, and initial versus continuing claims guide cover those neighboring measures.

Because JOLTS observes employers, it cannot show whether a particular unemployed person could access a particular vacancy or whether the job's pay, location, schedule, and required skills fit that person's situation. Openings and CPS unemployment can help describe matching conditions together, as in a Beveridge-curve comparison, but their aggregate counts do not identify individual matches or prove why hiring is slow. Treat a gap between vacancies and hires as a question to investigate with more evidence, not a direct count of unfilled jobs that a specific group of workers could take.

JOLTS is a sample survey of establishments, not a monthly census of every U.S. employer. The BLS FAQ says the survey samples approximately 21,000 units from a frame that covers more than 95% of nonfarm payroll jobs. That frame coverage does not make the monthly release a count of every job or every business. Published levels and rates are estimates built from establishment reports and survey procedures, so an initial release can be revised as more information arrives. A first monthly estimate is preliminary; the next month, the BLS incorporates additional respondent reports and recalculates seasonal factors. This is one reason to distinguish an early estimate from the later value in a revised history. The [JOLTS FAQ]({source:blsJoltsFaq}) describes the sample and release process.

The 95% figure describes how much payroll employment the sampling frame covers; it does not mean that 95% of vacancies are directly observed or that 95% of establishments respond. Approximately 21,000 refers to survey units, not jobs or openings. Remembering that the published level is estimated from sample reports helps keep it distinct from a complete count and puts month-to-month changes in the right context.

Annual updates can also revise the series. The BLS benchmarks JOLTS employment levels to updated CES employment estimates, which matter because employment is used in the denominators for JOLTS rates. The agency also reviews and updates seasonal-adjustment factors; its FAQ says the prior five years of seasonally adjusted and not-seasonally-adjusted data can be revised. A change in a historical rate therefore does not necessarily mean that the underlying opening, hire, or quit definition changed. It may reflect additional reports, an updated employment base, seasonal factors, or more than one of those inputs. The [JOLTS FAQ]({source:blsJoltsFaq}) describes the monthly and annual revision process.

When comparing a recent release with an older article, chart, or downloaded file, check whether both use the same data vintage and adjustment basis. For a short-term reading, use the latest revised series consistently across months; for research that must preserve what a reader could have known at the time, keep the original release vintage as well. The monthly headline remains useful, but its status as an estimate is part of what the number means.

For example, imagine that the first release shows openings falling from 5.40 million to 5.38 million. If the next release revises the earlier month to 5.35 million while the newer estimate remains 5.38 million, the updated comparison shows a 30,000 increase instead of the initial 20,000 decline. These are fictional values, not a claim about a typical revision. The example shows why a change in direction can come from a revised comparison month rather than a new definition of openings. When describing a move, identify whether the prior month was revised and use the same data vintage for both observations.

Think of the revision as a change to an estimate with a documented source, not as an extra month of hires or a change in the definition of an opening. If a chart uses seasonally adjusted values, compare it with another seasonally adjusted series; if it uses not-seasonally-adjusted values, keep that basis consistent. Labeling the release month and revision status makes an explanation easier to reproduce and prevents a preliminary value from being mistaken for the final estimate.