Three in ten, and one in two hundred
In 2025, 28.8% of Poland’s unemployed had been looking for twelve months or more, and 0.5% of South Korea’s: 57th and 106th of the 107 economies that record duration. Their unemployment rates are a quarter of a point apart, 3.0% and 2.8%, and both sit in the richest third of the set by income per person.
The spread is enormous. The median economy is Brazil at 30.2%; in 25 of the 107 more than half the unemployed have been looking a year or more, and in 9 fewer than one in ten. The highest figure is Oman’s, 86.9%, and the lowest the Philippines’, 0.2%.
The table holds both ends of that range. South Africa reports the highest unemployment rate of the seventeen, 32.3%, and the highest long-term share, 76.2%, while Italy is at 50.3% with a rate of 6.1%, and the United States at 13.8% with 4.3%.
| Economy | Year | A year or more, % of unemployed | Women | Men | Unemployment rate | Rank of 107 |
|---|---|---|---|---|---|---|
| South Africa | 2024 | 76.2% | 79.0% | 73.5% | 32.3% | 7 |
| Italy | 2025 | 50.3% | 50.1% | 50.5% | 6.1% | 25 |
| Portugal | 2025 | 36.8% | 36.6% | 37.1% | 6.0% | 40 |
| Indonesia | 2023 | 34.2% | 33.1% | 34.8% | 3.4% | 44 |
| Spain | 2025 | 32.1% | 34.0% | 30.0% | 10.5% | 48 |
| Brazil | 2025 | 30.2% | 34.8% | 25.0% | 5.8% | 54 |
| Poland | 2025 | 28.8% | 29.6% | 28.1% | 3.0% | 57 |
| Germany | 2025 | 27.9% | 25.2% | 29.9% | 3.8% | 59 |
| United Kingdom | 2025 | 23.5% | 20.0% | 26.3% | 4.9% | 67 |
| France | 2025 | 23.0% | 22.7% | 23.3% | 7.7% | 69 |
| Egypt | 2024 | 19.8% | 24.3% | 15.5% | 5.9% | 78 |
| Vietnam | 2024 | 19.4% | 21.8% | 17.5% | 1.5% | 80 |
| United States | 2025 | 13.8% | 12.7% | 14.7% | 4.3% | 94 |
| Netherlands | 2025 | 13.7% | 13.3% | 14.0% | 3.8% | 95 |
| Mexico | 2025 | 2.1% | 2.0% | 2.2% | 2.5% | 104 |
| South Korea | 2025 | 0.5% | 0.4% | 0.5% | 2.8% | 106 |
| Philippines | 2024 | 0.2% | 0.2% | 0.1% | 2.4% | 107 |
How many, and how long, are two questions
The two move together only loosely. Across the 107 economies, the long-term share correlates at 0.42 with the unemployment rate, which leaves most of the difference between any two economies unexplained: Poland and South Korea sit a quarter of a point apart on the rate and sixty times apart on the share.
Income explains less again. The share correlates at minus 0.28 with income per person, with medians of 43.2%, 32.0% and 23.4% from the poorest third of economies to the richest. The rate itself has no clean income gradient, correlating at minus 0.16, with medians of 3.9%, 5.8% and 4.9% across the same three groups.
Within the table the two figures come apart in both directions. Vietnam records the lowest unemployment rate of the seventeen, 1.5%, and still has 19.4% of its unemployed looking a year or more; France records 7.7% unemployment and 23.0%. Mexico, at 2.5% unemployment, is at 2.1%.
What a short spell can mean
A search ends in one of two ways, and this measure sees only one of them. A person who stops looking stops being unemployed, because the count requires active search and availability. A low long-term share is consistent with quick hiring, and equally with people leaving the search for study, family work, retirement or discouragement.
And every spell counted here is still running. Duration is time spent looking up to the survey week, measured on people who were still unemployed on that date, so the figure describes who is waiting now, not how long a completed search takes. A market where many people search briefly and give up can print a very low share.
Sex barely separates the two lines. Women’s long-term share is above men’s in 63 of the 107 economies, and the median gap between them is 0.8 points. In the table, Brazil is the widest, 34.8% for women against 25.0% for men.
- THE DENOMINATOR IS THE UNEMPLOYED. Not the labour force and not the population: this is the composition of one group, not its size. An economy can have few unemployed and nearly all of them long-term, or many unemployed and almost none.
- UNEMPLOYED MEANS SEARCHING AND AVAILABLE. The count holds only people who looked for work in the reference period and could start. Someone who stops looking leaves it, so a spell can end in a job or in leaving the search, and this measure cannot tell the two apart.
- EVERY SPELL HERE IS UNFINISHED. Duration is the time spent looking up to the survey week, counted on people who were still unemployed on that date. It is not the length of a search that ended, and no average waiting time can be read off it.
Poland’s 28.8% therefore does not say that a Polish job search takes a year, nor that South Korea’s 0.5% is a faster market: it says that of the people counted as unemployed in Poland in 2025, 28.8 in a hundred had been looking for twelve months or more, and in South Korea 0.5 in a hundred had.
One share, one denominator, and a reading that forgets who is being counted claims what the source does not say.
What the data establishes
- In 2025, 28.8% of Poland’s unemployed had been looking for twelve months or more and 0.5% of South Korea’s, 57th and 106th of 107 economies, with unemployment rates of 3.0% and 2.8%.
- Across the 107, the median long-term share is 30.2%, which is Brazil’s; 25 economies are above 50% and 9 under 10%.
- The long-term share correlates at 0.42 with the unemployment rate and at minus 0.28 with income per person, with medians of 43.2%, 32.0% and 23.4% from the poorest third to the richest; the unemployment rate itself correlates at minus 0.16 with income.
- Women’s long-term share is above men’s in 63 of the 107 economies, with a median gap of 0.8 points.
What it does not
- That a low share means a better labour market. A spell ends in a job or in leaving the search, and this measure cannot tell the two apart.
- How long a search takes. Every spell counted here is unfinished, so no average or median waiting time follows from it.
- Anything about unemployment benefits, their duration, training schemes or any other policy. None is measured here, and none is named as a cause.
- Any trend. One year per economy, with different years and instruments between them: nothing here says a share rose or fell.
- Anything about the 30 economies excluded for unrecorded duration, or about India, Japan and China, which report no duration at all.
Method
- Two measures, one survey. ILOSTAT DF_UNE_TUNE_SEX_AGE_DUR_NB, unemployment by sex, age and duration, and DF_UNE_DEAP_SEX_AGE_RT, the unemployment rate, ages 15 and over, both read through the ILO’s SDMX API on 22 September 2026 (the date is the timestamp of the downloaded files, 20:57 and 21:01). For each economy the latest year from 2019 to 2025 is used, and the duration counts, the rate and the figures for women and men all come from the SAME year AND the same reported source. The long-term share is the count of unemployed people with a duration of twelve months or more divided by all the unemployed of that year.
- AGGREGATES ARE NOT COUNTRIES, AND AN UNRECORDED DURATION IS NOT A SHORT ONE. Regions and income groups were removed by keeping only codes on the World Bank’s country list. Where a survey leaves the duration unclassified, the share would be diluted by a gap in the questionnaire rather than by short searches, so 30 economies with more than 5% unclassified are excluded and the largest are named here: Pakistan 89%, Samoa 88%, Burkina Faso 62%, Mozambique 44%, Timor-Leste 35%, Türkiye 20%, Thailand 7%. India, Japan and China report no duration at all. That leaves 107: 57 whose latest year is 2025, 20 with 2024, 7 with 2023, 13 with 2022, 7 with 2021, 1 with 2020 and 2 with 2019.
- THE INSTRUMENTS DIFFER. 93 of the 107 figures come from a labour force survey, 7 from a household income and expenditure survey, 5 from another household survey, 1 from administrative records and 1 from official estimates. Six carry a break-in-series flag, the United States and Egypt among them; a break matters when one year is set against another, and nothing here does that.
- The median of 107 values is the 54th, 30.241%, which is Brazil’s, shown as 30.2%. 25 economies are above 50%, 61 above 25%, 9 under 10%, 5 under 5% and 2 under 1%. Poland’s figure is 28.841% and South Korea’s 0.484%, a ratio of 59.5 before rounding; their unemployment rates are 3.018% and 2.789%.
- The correlations are Pearson coefficients over the 107, taken against the unemployment rate and against the natural log of GDP per person at purchasing power parity (World Bank NY.GDP.PCAP.PP.CD, latest year). The income thirds hold 35, 35 and 37 economies: the poorest below 14,300 dollars a head, the richest above 49,800. Each third’s figure is the median of its members. Poland’s income per person is 54,262 dollars and South Korea’s 63,125.
- Rounding is done once, at display, from figures the source serves to three decimals: South Africa 76.199%, Italy 50.289%, Germany 27.866%, the United Kingdom 23.462%, the Netherlands 13.673%, the Philippines 0.161%. The women’s and men’s shares are computed the same way, from the same year’s counts.
Primary sources
BabZituna scores every job against your profile on six real dimensions, and publishes what it can and cannot measure. See the live market.