Skip to content
Sign inGet started free
Photo: Ngcali Alexa Magopeni / Unsplash · South Africa

Market watch · 59

65.9% of young unemployed South Africans have never been employed

The same 3,088,152 people, counted a second way: 74.7% had been looking for twelve months or more.

South Africa’s quarterly labour force survey asks the unemployed two separate questions: whether they have ever been employed, and how long they have been looking. It publishes both answers for the same count of people aged 15 to 29, in two tables that are never crossed. Here is what each cut says, what happens to both of them as age rises, how the eleven published years refuse the word record, and what neither table lets anybody conclude.

65.9%of the unemployed aged 15 to 29 were seeking their first job
34.1%had held a job before, the rest of the very same count
74.7%of that same count had been looking for twelve months or more

The share with no previous job falls with age

15 to 1988.6%
20 to 2476.0%
25 to 2954.2%
Unemployed people seeking their first job, percent of all unemployed people in the band, South Africa, 2025, for the three age bands the table publishes. Each bar is one band’s published first job cell over that band’s published total, both counts of people and not rates: the unemployment rate for the same bands is a share of the labour force and is printed in the table below. The bands are three different groups of people counted in the same year, so the fall across them is a cross-section and not anybody ageing. ILOSTAT, youth unemployment by sex, age and categories of unemployed persons, indicator UNE_3UNE_SEX_AGE_CAT_NB, from the Quarterly Labour Force Survey, read 11 November 2026.

One count, two questions

In 2025 South Africa counted 3,088,152 people aged 15 to 29 as unemployed. Every one of them was without work, available for work and looking for it, which is what the word counts in this survey, and the figure is the annual reading of the country’s quarterly labour force survey.

The survey then asks those people two separate questions, and publishes both answers for the same count. The first is whether they have ever been employed. The second is how long they have been looking. Each answer fills its own table, and the two tables add to the same total.

Neither table crosses the other. There is no published cell for people who had never been employed and had been looking a year, so this edition does not say how far the two groups overlap. What it can say is how large each is.

Provenance: SourceWhat the survey counts
  1. UNEMPLOYED here means without work, available for work and looking for work, all three at once. Everybody in this count was looking, which is why none of them can be described as idle, inactive or outside the labour force: a person who is not looking is outside the labour force and is not in this count at all.
  2. SEEKING A FIRST JOB and PREVIOUSLY EMPLOYED are the two categories of the first cut, a classification by previous employment, and together they are the whole of the count: 2,036,412 and 1,051,740 add to 3,088,152, and that addition holds in all 44 cells read, to within one person in the published thousandths.
  3. THE DURATION IS THE SEARCH SO FAR. The second cut asks people who are unemployed NOW how long they have been looking, so every class describes a spell in progress rather than a completed one. 74.7% having looked for twelve months or more is therefore a statement about how long the current search has run, and nothing about how long it will run or how it ends.
  4. TWO CUTS OF ONE COUNT, NEVER CROSSED. Both tables publish the same total for the same people, in all 44 cells read, which is what makes them two views and not two populations. But no published cell crosses them, so how many of the 2,036,412 first job seekers were among the 2,307,991 who had looked for a year or more is not in this data, and this edition does not estimate it.
  5. A SHARE OF THE UNEMPLOYED IS NOT AN UNEMPLOYMENT RATE. 65.9% is 2,036,412 divided by 3,088,152, both of them unemployed people. The unemployment rate divides the unemployed by the LABOUR FORCE, which is the employed plus the unemployed, and for these bands it reads 50.7%, 53.7% and 40.9% against 29.6% for everybody aged 15 and over. The two are never swapped, added or subtracted.
  6. AND THE AGE BANDS ARE A CROSS-SECTION, NOT A COHORT. The three bands are three different groups of people counted in the same year. The 25 to 29 band is not the 15 to 19 band ten years later, nobody is followed between bands or between years, and both tables stop at 29, so neither says anything at all about the unemployed aged 30 and over.

So these tables say how many young South Africans were looking for work in 2025, how many of them had ever been employed and how long they had been looking, and nothing about why, about what any job pays, about how any search ended, or about anybody aged 30 and over.

One count of people, two questions asked of them, and a third measure with another denominator entirely. A reading that crosses the two cuts, or that treats a share of the unemployed as an unemployment rate, says what these tables do not.

Two thirds have never been employed

2,036,412 of the 3,088,152, which is 65.9%, were seeking their first job. The other 1,051,740, 34.1%, had held one before. The two categories are the whole of the count: there is no third category and no unclassified remainder in that cut.

Read the same count the other way and 2,307,991 of them, 74.7%, had been looking for twelve months or more. 415,986 had been looking for less than six months, 345,394 between six and twelve, and 18,781 could not be classified by duration at all.

The two readings describe the same people and must not be joined. A reader who concludes that the first job seekers are the long waiting ones is adding a cell that the survey does not publish. Both facts are about one count of 3,088,152; neither is about a subgroup of the other.

Provenance: Source
One count of unemployed people aged 15 to 29 in South Africa in 2025, cut two ways, people and percent of the count. The first two rows are the first cut, by previous employment, and the four rows below them are the second cut, by how long the search has run. Both cuts describe the SAME 3,088,152 people and both add to that total, which is the control that makes them two cuts of one count rather than two counts: across the eleven years and four bands read, the first cut’s total and the second cut’s total are the same number in all 44 cells. The two cuts are never crossed in the published tables, so no row here says how many of the first job seekers were also the ones looking for a year or more. ILOSTAT, indicators UNE_3UNE_SEX_AGE_CAT_NB and UNE_3UNE_SEX_AGE_DUR_NB, from the Quarterly Labour Force Survey.
Cut of the same countPeoplePercent of the count
Seeking their first job2,036,41265.9%
Previously employed1,051,74034.1%
Looking for less than six months415,98613.5%
Looking for six months to less than twelve345,39411.2%
Looking for twelve months or more2,307,99174.7%
Duration not classified18,7810.6%

The two cuts cross over with age

The share seeking a first job falls steeply with age: 88.6% at 15 to 19, 76.0% at 20 to 24, and 54.2% at 25 to 29. Even in the oldest of the three bands, more than half of the unemployed had never held a job, which is the figure this edition finds hardest to read as a transition out of school.

The share looking for twelve months or more runs the other way: 58.1%, then 74.8%, then 77.0%. So the two cuts of one count move in opposite directions across the bands, and the band with the fewest first time seekers is the band with the longest searches.

The unemployment rate is a third measure with a third denominator, and it follows neither pattern. It reads 50.7% at 15 to 19, 53.7% at 20 to 24 and 40.9% at 25 to 29, against 29.6% for everybody aged 15 and over. These three bands are three different groups of people counted in the same year, so nothing in the fall across them is anybody growing older.

Provenance: Source
The two cuts by age band, South Africa, 2025. The first column is the published count of unemployed people in the band; the next two are that band’s first job cell and its twelve months or more cell as percentages of it, both shares of the UNEMPLOYED. The last column is a different measure with a different denominator, the unemployment rate, which is a share of the LABOUR FORCE and comes from a third indicator; it is printed here because the two must not be confused and is left empty for the whole range, for which no rate is published. The three narrow bands add to the wide one in both cuts, in all 77 cells read. The bands are a cross-section: three different groups of people counted in the same year, nobody followed. ILOSTAT, indicators UNE_3UNE_SEX_AGE_CAT_NB, UNE_3UNE_SEX_AGE_DUR_NB and UNE_DEAP_SEX_AGE_RT.
Age bandUnemployedSeeking a first jobLooking twelve months or moreUnemployment rate
Aged 15 to 19216,17588.6%58.1%50.7%
Aged 20 to 241,323,58676.0%74.8%53.7%
Aged 25 to 291,548,39154.2%77.0%40.9%
Aged 15 to 293,088,15265.9%74.7%No data

Eleven years, and not a record

Both shares are higher than they were a decade ago. The first job share was 58.1% in 2015 and 65.9% in 2025, a rise of 7.8 points, and the twelve month share was 62.6% against 74.7%, a rise of 12.1 points. Over the same eleven years the count itself rose by 490,338, from 2,597,814 to 3,088,152.

Neither share is at its highest, and this edition does not call either a record. 65.9% is the fourth highest first job share of the eleven years, behind 69.9% in 2022, 66.3% in 2023 and 66.1% in 2021. 74.7% is the third highest duration share, behind 76.1% in 2021 and 75.5% in 2022.

And the count fell in the latest year, by 296,726 from 3,384,878 in 2024. So the count and the shares did not move together, and no sentence here describes them with one verb. The eleven years are printed in full for that reason: a single year would have allowed either claim.

Provenance: Source
Unemployed people aged 15 to 29 in South Africa, and the two cuts as percentages of that count, for each of the eleven published years. The table is here because one year could be a year picked to suit the finding, and it refuses two readings: 65.9% is the fourth highest first job share of the eleven, 2022 holding the highest at 69.9%, and 74.7% is the third highest duration share, 2021 holding 76.1%. It also shows the count itself falling in the latest year while both shares held, which is why the count and the shares are never described with one verb. ILOSTAT, indicators UNE_3UNE_SEX_AGE_CAT_NB and UNE_3UNE_SEX_AGE_DUR_NB, annual figures from the Quarterly Labour Force Survey.
YearUnemployed aged 15 to 29Seeking a first jobLooking twelve months or more
20152,597,81458.1%62.6%
20162,753,71460.7%63.8%
20172,830,36960.5%64.5%
20182,803,26161.9%67.2%
20193,010,02261.3%68.4%
20202,664,01463.0%70.3%
20213,193,96566.1%76.1%
20223,274,49669.9%75.5%
20233,303,09566.3%73.9%
20243,384,87865.5%73.9%
20253,088,15265.9%74.7%

What these figures do not say

They do not say that the first job seekers are the long waiting ones. The two cuts are published in two tables that share a total and never cross, so the overlap between the 2,036,412 who had never been employed and the 2,307,991 who had been looking for twelve months or more is not in this data. Nor do they say that people who had held a job find work faster: the survey measures the search in progress, never its outcome.

They do not say that either share is a record, or that youth unemployment rose. 65.9% is the fourth highest of the eleven published years and 74.7% the third, and the count of unemployed people aged 15 to 29 fell by 296,726 between 2024 and 2025. A reading in which everything worsened at once is not what the eleven years show.

They do not say that anybody counted here is idle or has stopped looking. To be in this count a person had to be looking for work, which is the test the survey applies; those who are not looking are outside the labour force and outside these tables entirely. And they do not say why any of these shares is what it is: the survey records labour market status, never the reason, so this edition names no cause, not schooling, not training, not hiring rules, not wages, not policy and not the economy.

And they are two youth tables, for one country, for one year. Both stop at 29, so nothing here concerns the unemployed aged 30 and over; nobody is followed between bands or years, so a difference between bands is a difference between groups of people; nothing is broken out by province, sector, occupation, education or population group, none of which was read; no other country appears, because none was read; and nothing in this edition is a forecast or comes from our own platform.

What this means for you

If you are hiring in South Africa

Two thirds of the young people available to you have no previous employer, and a requirement of prior experience therefore screens out 2,036,412 of the 3,088,152 unemployed aged 15 to 29 before anything else is considered. The share is 88.6% in the 15 to 19 band, 76.0% at 20 to 24 and still 54.2% at 25 to 29, so it is not confined to school leavers. The second cut of the same count says that 2,307,991 of these people, 74.7%, had been looking for twelve months or more, which is the group a filter on time since the last job would remove. The two cuts are never crossed in the published tables, so how much those two groups overlap is not in this data. Nothing here is broken out by province, sector, occupation or education, none of which was read, and nothing in it says why any share is what it is.

If you are looking for work in South Africa

If you have never been employed you are in the larger of the two groups the survey counts, not the smaller: 2,036,412 people aged 15 to 29 were seeking a first job in 2025 against 1,051,740 who had held one. The unemployment rate for your band is a different measure again, a share of the labour force rather than of the unemployed: 50.7% at 15 to 19, 53.7% at 20 to 24 and 40.9% at 25 to 29, against 29.6% for everybody aged 15 and over. The duration figures describe searches in progress and not finished ones, so 74.7% having looked for twelve months or more says how long the current search has run, never how long it will run or how it ends. And the count of young unemployed people fell by 296,726 between 2024 and 2025 while both shares stayed near their decade highs, so the two move separately.

How this was counted

Method

  1. THE READ. Three indicators, read on 11 November 2026 at 21:30 UTC from ILOSTAT’s keyless data interface at rplumber.ilo.org, each as comma separated values with labels: UNE_3UNE_SEX_AGE_CAT_NB, youth unemployment by sex, age and categories of unemployed persons, in thousands; UNE_3UNE_SEX_AGE_DUR_NB, youth unemployment by sex, age and duration, in thousands; and UNE_DEAP_SEX_AGE_RT, the unemployment rate by sex and age, in percent. All three were requested for South Africa from 2015 onwards. The two youth tables publish four age bands, 15 to 19, 20 to 24, 25 to 29 and the whole 15 to 29 range, and eleven annual readings ending in 2025, which is the latest they hold.
  2. WHAT THE DATA SAYS ABOUT ITSELF. Every row of all three responses carries the survey behind it, "LFS - Quarterly Labour Force Survey", and the repository that processed it, "ILO-STATISTICS - Micro data processing". The response does not name the office that runs that survey, so this edition cites ILOSTAT and the survey rather than an office, and the figures are annual readings of that quarterly survey rather than a single quarter. That provenance is therefore in the data rather than on a definitions page, which matters because it is what names the survey whose categories this edition turns on. The counts are published in thousands to three decimals, so a cell of 3.088152 thousand thousands is 3,088,152 people: the whole people printed here are the publication’s precision, not the survey’s, and they remain weighted estimates from a sample.
  3. THE CONTROLS. Four identities were run over the saved reads before anything was written, and all four passed. The two categories of the first cut add to the published total, in 44 cells. The three narrow bands add to the wide band, in both cuts, across 77 published series. The four classes of the second cut add to the published total, in 43 cells, the one remaining band and year being the single case where the unclassified cell is withheld. And the first cut’s total equals the second cut’s total, in all 44 cells, exactly: that last one is what makes them two cuts of one count. The largest gap any of the other three leaves is one person in the published thousandths.
  4. THE FLAGS. 53 cells of the duration table as read carry a low reliability flag, and every one of them falls on the class that could not be classified by duration. Not one cell of the first cut carries a flag or is withheld. Of the figures printed in this edition, none is a flagged cell: the one flagged cell inside the printed range is the unclassified count for the 15 to 19 band, which is why that cell appears in no table here and the unclassified figure is given only for the whole 15 to 29 range.
  5. DERIVED FIGURES. The counts are published; every percentage in this edition is this edition’s own division of two published counts, and every point change is the difference of the two one decimal percentages printed beside it, so a reader repeating the subtraction lands on the same answer. Computed that way: the shares of 65.9%, 34.1%, 74.7%, 88.6%, 76.0%, 54.2%, 58.1%, 74.8% and 77.0%; the 2015 shares of 58.1% and 62.6% and the decade rises of 7.8 and 12.1 points; the highest shares of 69.9% and 76.1% and the ranks of fourth and third; and the count changes of 490,338, 528,380 and 296,726, which are subtractions of published counts and therefore exact. The unemployment rates of 50.7%, 53.7%, 40.9%, 29.6%, 53.2% and 26.6% are published rather than derived.
  6. WHAT WAS TESTED AND REFUSED. Three claims were checked against the eleven years and dropped. That the first job share is at a record: it is the fourth highest of the eleven, behind 2022, 2023 and 2021, and the duration share is the third highest, behind 2021 and 2022, which is why the whole series is printed rather than the latest year alone. That youth unemployment rose: the count fell by 296,726 between 2024 and 2025 even as both shares stayed high. And that the people seeking a first job are the people who had been looking longest, which is the most natural reading of the two cuts together and cannot be made from them at all, no published cell crossing the two tables. What did survive is the crossover across the age bands, which the band table prints.
  7. LIMITS. Nothing here names a cause, follows a person, measures pay, hours, occupation, sector, education, province or population group, says anything about the unemployed aged 30 and over, crosses the two cuts, compares South Africa with another country, or reaches outside the eleven published years. A share of the unemployed is never read as an unemployment rate, the duration classes are never read as outcomes, and no figure in this edition comes from our own platform.
What the data establishes, and what it does not

What the data establishes

  • In 2025, 3,088,152 people aged 15 to 29 in South Africa were counted as unemployed, of whom 2,036,412, which is 65.9%, were seeking their first job and 1,051,740, 34.1%, had held one before.
  • The same count read by duration says that 2,307,991 of those people, 74.7%, had been looking for twelve months or more, and that the two cuts add to the same total in all 44 cells read.
  • The share seeking a first job falls across the age bands, 88.6% at 15 to 19, 76.0% at 20 to 24 and 54.2% at 25 to 29, while the share looking for twelve months or more rises, 58.1%, 74.8% and 77.0%.
  • Over the eleven published years the first job share rose from 58.1% to 65.9% and the duration share from 62.6% to 74.7%, while the count rose by 490,338 from 2,597,814.
  • Neither share is at its highest: 65.9% is the fourth highest first job share of the eleven years and 74.7% the third highest duration share, and the count fell by 296,726 from 2024 to 2025.

What it does not

  • That the people seeking a first job are the people who had been looking longest, or that those who had held a job find work faster: the two cuts are never crossed and the survey records no outcome.
  • That either share is a record: 65.9% is the fourth highest of the eleven published years, behind 69.9% in 2022, and 74.7% is the third highest, behind 76.1% in 2021.
  • That youth unemployment rose in 2025: the count of unemployed people aged 15 to 29 fell by 296,726 from 3,384,878 in 2024 to 3,088,152.
  • That anybody in this count is idle, inactive or outside the labour force: every person in it was without work, available for work and looking for work, which is the test the survey applies.
  • That 65.9% is an unemployment rate: it is a share of the unemployed, while the rate is a share of the labour force and reads 50.7%, 53.7% and 40.9% in the three bands.
  • Why any figure is what it is: the survey records labour market status and never the reason, and this edition names no cause, schooling, training, employers’ hiring rules, wages, policy and the economy included.
  • Anything about the unemployed aged 30 and over, who are outside both tables, and anything about pay, hours, occupation, sector, education, province or population group, another country, or any year to come.
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.

The markets in this piece, live

Earlier editions