EN

Unemployment rate calculator (ILO definition)

The unemployment rate isn't the share of unemployed people in the total population: it's their share of the labor force — only the people who are working or actively looking for work. A student, a retiree, or a stay-at-home parent are neither employed nor unemployed: they're out of the labor force, and appear in neither term of the ratio.

That shift in denominator is behind half of all calculation errors, and it's why a country can post a low unemployment rate alongside a mediocre employment rate. This calculator returns all four rates together — unemployment, participation, employment, and the rate of people out of the labor force — plus the broader rate that counts people who want work but aren't formally counted as unemployed.

Someone counts as unemployed under the ILO definition only if they meet all three conditions at once: without work, available to work, and actively searching. That's not the same as being registered with an employment agency — the two counts differ structurally, in both directions.

Labor market figures

Labor force

Everyone with a job: employees, self-employed, apprentices, part-time included.

Without work, available, and actively searching.

Figures in whatever unit you like — headcount or thousands — as long as it's the same throughout: the rates are ratios.

Beyond the unemployment rate alone (optional)

Conventionally ages 15-64. Needed for the participation and employment rates.

People out of the labor force who want a job but don't meet all three ILO criteria.

marginal 1,900employedinactiveunemployed 2,28028,50011,020labor force A = E + Uworking-age population Punemployment rate = U / A =7.41%employment rate = E / P = 68.18%
Worked example (demonstration values): the unemployment rate refers to the lower bracket (the labor force), the participation and employment rates to the upper bracket (the whole working-age population). The figure redraws with the values entered once calculated.

Asterisks mark the labour-market jargon; each one is unpacked in the glossary below the article.

Four boxes, not two: employed, unemployed, inactive, working-age

Working-age people sit in four boxes, not two. The unemployment rate is a ratio of two of them. Mix the boxes and the percentage is theatre.

The labor force* is everyone who is either in a job or meeting the unemployment tests. Students, retirees, carers, the long-term sick, people who have stopped looking: they are inactive* (out of the labor force). They belong in neither the numerator nor the denominator of U / A. A stay-at-home parent is not “unemployed” in this arithmetic. Neither is a 63-year-old on a pension.

A = E + U  ·  unemployment rate = U / A

participation rate = A / P  ·  employment rate = E / P

P is the working-age population, often 16–64 in UK tables, 15–64 in many international ones. Move 100,000 people from “unemployed” to “inactive” and the rate falls even though nobody was hired. That boundary is the whole trick. Fill in P on this page and the employment rate* sits next to the unemployment rate, which is the pair that actually describes a market.

The ILO’s three tests, and why one hour is enough

The ILO* (International Labour Organization) does not count hardship. It counts people who pass three tests in the same reference week, all of them.

  1. Without work — not even one hour in that week, paid or unpaid family work above the survey threshold.
  2. Available — able to start within two weeks.
  3. Actively seeking — at least one active step in the past four weeks, or a job already lined up that has not started yet. Scrolling adverts without applying does not always count.

Miss any one gate and the person is not unemployed. They are inactive — even if they want work, even if they have given up, even if they are waiting on a course or on childcare. Restrictive on purpose. The payoff is comparability: a rate in Manchester and a rate in Milan can share a definition. The cost is a thinner unemployment total than the hardship people actually feel. The U in this form is the ILO headcount, not a benefit caseload.

Labour Force Survey versus the claimant count

The number a Monday paper calls “unemployment” is often the claimant count*, not the ILO rate. Two machines, two jobs.

The ILO count comes from a household survey. In Britain that is the ONS* (Office for National Statistics) Labour Force Survey: interviewers apply the three tests. It is the series for a cross-country chart or a macro diagnosis. The claimant count is a DWP caseload — people on Jobseeker’s Allowance plus those on Universal Credit required to search for work. Jobcentre Plus, and the online UC journal that has replaced a lot of old signing-on, is the shop floor of that caseload. Some claimants are already in work, with a taper. Some sit in a “no work requirements” group and never belonged in a jobseeker total. Some ILO-unemployed people never register at all.

Dropping a claimant-count figure into this calculator as U only works if those people meet the three ILO tests. Most of the time they do not, not one-for-one. Use LFS (or ONS ILO) unemployed for U, and keep the claimant count in a separate sentence.

Universal Credit scrambled the comparison with history. UC is one payment for several old benefits, and the “searching for work” conditionality does not map onto JSA one-for-one. A headline that says “unemployment up” because UC claims rose during a cost-of-living shock may be counting people who already had a job and needed help with rent.

Marginally attached workers and the broader rate

Around the official unemployed sits a halo of people who want work but fail a test. US statisticians call them marginally attached*. The UK talks about a “want work” group among the inactive, and about discouraged workers as a subset.

  • Want a job, not actively searching — discouraged, waiting for a course, stuck on childcare, waiting for a fit note.
  • Searching, but not available within two weeks.

They are not in U. They are in the inactive slice. Add them back and the form prints (U + H) / (A + H). The halo swells in a long slump and shrinks when hiring is easy. Underemployed part-timers are a different halo again — they already sit in E — and should not be dumped into H without changing the story.

A two-point gap between the official rate and the broader one is not a rounding error. It is hidden unemployment with a name.

Underemployment: hours that never came back

A job can be too small anyway. Underemployment is hours, not heads: people in work who want more hours and are available to do them, plus people on reduced hours for lack of demand (short-time, a pub on three days, a factory on a three-day week).

They are employed. They do not raise U. They do raise a different ratio, and they are why a recovery can post a falling unemployment rate while living standards go on feeling like a recession. Involuntary part-time is the British version in ONS labour-market releases. Zero-hours contracts complicate the picture further: the person is employed, sometimes for a handful of hours, and may be looking for something else without leaving E.

U / A watched alone makes the market look healed the week the hours have not come back. Hours and the underemployment count belong on the same desk.

Low unemployment with weak employment: inactivity doing the work

A low unemployment rate with a poor employment rate is not a paradox. It is arithmetic. Employment rate is E / P. Participation is A / P. Unemployment is U / A. Pull people out of A — more students, more early retirees, more long-term sick, more carers — and U / A can look tidy while E / P is nothing to boast about.

Southern Europe has lived this with young people and older men. The UK has had stretches of it too, especially when inactivity among the 50–64s rose after the pandemic. The unemployed denominator shrank. The working-age population did not. The chart to pin on the wall is the employment rate next to the unemployment rate, not the unemployment rate alone. This form prints both once P is filled in.

The unemployment rate also says nothing about how unequally pay is shared once people are in work. That is a different instrument — the Gini coefficient on incomes, not a headcount of jobless people.

Discouraged workers: a tidier rate that is not a recovery

Recession drags on. Applications go nowhere. Some of the unemployed stop searching in a way the survey will count. They fail test three. They leave U and enter inactivity. A falls, U falls faster, U / A drops. Nobody hired them. The market did not improve. The statistic got tidier because the definition pushed them off the page.

The reverse happens in a boom: people who had given up start looking again, A swells, and the unemployment rate can tick up just as hiring improves. That is why a single month’s rate, read in isolation, is a trap. Pair it with participation and with the employment rate, every time. In practice the “good news” unemployment fall that coincides with a jump in inactivity is the discouraged-worker effect wearing a press-release smile.

Duration, age, sex, education: one average, several markets

The headline rate is an average of very different rooms.

  • Long-term — twelve months or more of search. In Europe it is often around a third of the unemployed in ordinary years, and can climb toward 40–50% after a long crisis. Skills rust. Employers flinch at the gap. Getting this group back is a different policy from filling a three-week vacancy.
  • Age — youth unemployment (15–24) typically runs two to three times the rate for ages 25–49. Older workers (50+) often show a lower rate but longer spells once they are in. Youth rates have a small denominator: many are in education, so each unemployed person hits hard. The NEET share (not in employment, education or training) is often the better youth distress signal than U / A.
  • Sex — the gap is a country story. US rates often sit close together; elsewhere women carry more of it, or less, depending on participation as much as on layoffs.
  • Education — lower qualifications, higher unemployment, even when the headline rate looks like full employment. The residual is stubborn.

A national rate of 4% can hide 12% for one age-education cell. Policy that only watches the average will train the wrong people.

Frictional, structural, cyclical: three diagnoses, three tools

Not all unemployment is a missing factory order. Economists split it because the medicine differs.

  • Frictional — time to search and match. Always there, even at “full employment.” The natural rate is not zero. Better job boards shrink it a bit; they do not abolish it.
  • Structural — skills, place, industry. A closed pit, a coastal town, a software stack nobody over 50 was taught. Training, mobility, industrial policy. Demand stimulus will not retrain a welder into a care worker by itself.
  • Cyclical — the slump. Output below potential. This is the slice fiscal and monetary stimulus are for.

Train people in a demand collapse and certificates pile up. Pump demand into a skills mismatch and vacancies sit next to unemployment in the same month. Worth noting: the three add up in the data only as a story, not as three published lines available for download. The mix is inferred from vacancies, duration, regional spreads, and how the rate moves with GDP.

Seasonally adjusted prints and the next revision

July is not January. School leavers, Christmas temps, harvests, tourism. Raw unemployment jumps for reasons that repeat every year. Seasonally adjusted series strip that calendar out so the trend is visible. Compare a raw July with a seasonally adjusted March and a story has been invented.

Then the revisions. The BLS, Eurostat, the ONS — they all recast. New interviews arrive, seasonal factors are re-estimated, a method changes. Levels move. Sometimes the slope of the last eighteen months flips sign. That is normal. It is also why a Monday headline built on the first print is a hostage. Last year’s “sharp fall” may this year be a mild one, same months, new vintage.

SA is the series for the cycle. Raw is the series for a summer hiring desk. Splicing them in one chart without a footnote invents a turning point.

Okun’s law as a back-of-envelope check

Arthur Okun noticed a stubborn regularity: unemployment moves with the output gap. A common classroom version is

ΔUnemployment ≈ −0.5 × (GDP growth − potential growth)

Grow at potential and the rate holds. Grow faster, and it eases; slower, and it rises. In the United States the coefficient is often around −0.5: an extra point of growth, about half a point off unemployment. In Europe it is sometimes smaller, with more structural stickiness in the way. It is a rule of thumb, not a law of physics — participation shifts, hours shifts, and productivity bursts all break it for a year or two.

As a back-of-the-envelope check it is hard to beat. GDP up 3%, potential 1.5%, coefficient −0.5: unemployment down by something like 0.75 points would fit. If the rate has barely moved, inactivity and hours are the next place to look, before throwing the relationship away.

A cross-country chart that does not mix definitions

A UK rate next to a US rate next to a Spanish rate is a minefield. Walk it like this:

  • Definition — ILO survey, or an administrative register?
  • Age — 15–64, 16–64, 15–74, 16+? Youth rates are hypersensitive to whether students sit in P.
  • Period — quarterly snapshot, three-month rolling average (the ONS habit), annual mean.
  • Adjustment — SA or not.

Eurostat’s harmonised unemployment rates put European countries on the ILO footing. The OECD does the same for a wider club. National claimant series do not belong on that chart. Even ILO-on-ILO can wobble: survey design, the treatment of conscripts, and how “one hour” is asked all leave fingerprints. The methodological note is dull, and cheaper than a wrong league table.

This calculator does not harmonise anything. It divides the stocks it is given. The ILO criterion lives in how E and U were classified, not in the button.

U-3 and U-6: the BLS ladder of slack

The US Bureau of Labor Statistics publishes a ladder, not a single rate. U-3 is the official unemployment rate — the ILO-style U / A that this page computes. U-6* is broader: unemployed, plus marginally attached, plus people employed part-time for economic reasons, all over a labor force that has been widened by the marginally attached. In a deep slump U-6 can sit at roughly twice U-3. In a tight market the gap narrows; it rarely vanishes.

U-6 is not “the real unemployment rate.” It is a different question: how much unused labor is sitting in and around the official unemployed. Eurostat’s “labour underutilisation” family is the European cousin, not a photocopy. Feeding involuntary part-time into H on this form, without changing the story, produces a hybrid that is neither U-3 nor U-6.

The NAIRU* (non-accelerating inflation rate of unemployment) sits in a different drawer again: an estimated floor below which inflation tends to pick up. It is not a BLS release. It is a model object, with error bands of a point or more, and it is a poor overlay on a country whose U-6 and U-3 tell two different tightness stories.

ONS Labour Force Survey: sample, pause, and the 2020s revisions

UK ILO figures come from the Labour Force Survey: a household sample, interviews, the three tests applied to each adult. For years it was the unchallenged source. Response rates then fell, the sample tilted, and in the mid-2020s the ONS had to warn users, reweight, and — for a stretch — lean on other sources while it rebuilt. Seasonally adjusted rolling three-month estimates were revised, sometimes sharply. Employment and inactivity moved as much as unemployment.

That is not a scandal unique to Britain. It is what a voluntary household survey looks like when people stop answering the phone. The rates typed into this form are only as good as the survey that produced E, U and P. A 0.3-point “rise” inside a year of reweighting may be noise. Recent UK LFS prints deserve a wider interval than a textbook would, and a cross-check against payrolls (PAYE) and the claimant count, without pretending they measure the same thing.

Furlough in 2020: employed on paper, hours gone

The Coronavirus Job Retention Scheme paid firms to keep people on the books at 80% of wages, later tapered. Those people had no hours. Many of them counted as employed under the ILO all the same, because they had a job to return to. Unemployment rose far less than GDP collapsed. Employment stayed on the books. Hours fell off a cliff.

So the unemployment rate, for a year, was the wrong instrument. Inactivity moved. Underemployment and hours told the truth. When furlough ended, some of those jobs vanished and U ticked up; some returned to hours and the rate barely shrugged. Another scheme that pays people not to work while keeping the contract alive will do the same: headcount employment in this form will look healthy while the hours are not.

In short, U / A is a clean ratio with a fussy numerator and a fussy denominator. Both need a name, so does the survey, and the claimant count belongs in its own box.

Glossary

Labor force
Everyone in a job plus everyone who meets the three unemployment tests. Students, retirees and carers sit outside it. The unemployment rate divides the unemployed by this group, not by the whole population.
Inactive / out of the labor force
Working-age people who are neither employed nor unemployed under the ILO tests. They want work or they do not; either way they are missing from both sides of U / A.
ILO
International Labour Organization. The UN agency whose three tests (without work, available, actively seeking) define official unemployment for cross-country tables.
Claimant count
UK administrative caseload: people on Jobseeker’s Allowance plus Universal Credit claimants required to search for work. A DWP register, not an ILO headcount. The two move differently.
ONS
Office for National Statistics. Runs the UK Labour Force Survey that applies the ILO tests to a household sample. That survey, not the claimant count, is the ILO series.
Marginally attached
People out of the labor force who want a job but fail a test — not searching, or not available within two weeks. The US label; the UK talks about a “want work” group among the inactive.
Employment rate
Employed people as a share of the working-age population (E / P), not of the labor force. A tidy unemployment rate can sit next to a poor employment rate if inactivity is high.
U-6
A broader US slack measure from the BLS: unemployed, plus marginally attached, plus involuntary part-time, over a labor force widened by the marginally attached. U-3 is the official U / A on this page.
NAIRU
Non-accelerating inflation rate of unemployment: an estimated floor below which inflation tends to pick up. A model object with wide error bands, not a survey print.

Related calculators