People
Korea's listed pop-culture companies employ 15,072 people. They stay half as long as the rest of the market.
Netflix's Korean catalogue has been watched 23.7 billion hours. The listed companies in that business keep staff 5.5 years against 11.3 across Korea's listed market — and their two halves are mirror-image workforces.
Korea’s pop-culture export is usually measured by what the world watched. Netflix has published its weekly global Top 10 for 265 straight weeks, and over that span 236 Korean titles have appeared on it, for 23.7 billion hours viewed. Squid Game alone accounts for 5.05 billion of those hours and held the chart for 42 weeks.
The other side of that trade — the people — is almost never counted. It can be. Korean companies disclose the average tenure, headcount, pay and gender split of their own staff in their annual filings. Fourteen listed entertainment and game companies file that disclosure. Between them they employ 15,072 people.
They leave in half the time
The average employee at these 14 companies has been there 5.5 years, weighting each company by how many people it employs. The same field on the same filing, across all 2,779 listed Korean companies that report it, gives 11.3 years for 1.87 million workers.
So the industry that carries Korea’s cultural export turns its staff over roughly twice as fast as the listed economy around it.
The obvious explanation is age: these are young companies, and you cannot have eleven-year tenure at a ten-year-old firm. It does not hold up on its own. These 14 average 26 years since incorporation against 43.5 for the listed market — younger, but not half as young, while tenure is roughly half. Within the 14, tenure and company age correlate at only 0.30, and the ratio between them runs from 13% to 46% with no visible pattern. Youth is part of it. It is not the whole of it, and we cannot say from a filing what the rest is.
Two halves, mirror images
Split the 14 by the industry code they file under and the tenure gap between entertainment and games nearly disappears — 5.3 years against 5.6. Almost everything else about the two workforces is opposite.
| Company | Staff | Tenure (yr) | Women | Avg pay (₩m) | Age (yr) | |
|---|---|---|---|---|---|---|
| Entertainment and production | 6,186 | 5.3 | 65.1% | 81.9 | 28.9 | |
| CJ ENM | 3,105 | 6.89 | 63.0% | 83.8 | 31.7 | |
| HYBE | 893 | 2.96 | 68.3% | 93.3 | 21.5 | |
| SM Entertainment | 765 | 4.10 | 71.8% | 70.1 | 31.5 | |
| JYP Entertainment | 496 | 4.21 | 70.6% | 96.3 | 30.3 | |
| YG Entertainment | 438 | 3.95 | 65.8% | 61.0 | 28.5 | |
| Studio Dragon | 179 | 4.76 | 60.3% | 94.1 | 10.3 | |
| Cube Entertainment | 170 | 2.19 | 65.3% | 52.9 | 13.1 | |
| Contentree JoongAng | 140 | 5.23 | 37.9% | 67.6 | 38.9 | |
| Games | 8,886 | 5.6 | 33.9% | 108.6 | 23.5 | |
| NCSoft | 3,262 | 7.47 | 32.8% | 116.9 | 29.4 | |
| Krafton | 2,118 | 3.20 | 34.9% | 128.9 | 19.4 | |
| Com2uS | 1,505 | 5.10 | 34.8% | 81.4 | 28.0 | |
| Netmarble | 795 | 5.91 | 36.1% | 87.4 | 14.8 | |
| Pearl Abyss | 733 | 4.54 | 28.9% | 103.6 | 15.9 | |
| Kakao Games | 473 | 5.33 | 38.5% | 90.2 | 13.0 |
Seven of the eight entertainment companies are majority-women, from 60.3% to 71.8%. All six game companies are majority-men, from 61.5% to 71.1%. The ranges do not overlap. The single exception is Contentree JoongAng at 37.9%, which sits inside the games band rather than beside its own group.
Games pay more: ₩108.6m per employee against ₩81.9m, about a third higher on the same disclosure.
Put plainly, the K-pop and drama business is a workforce of women that is paid less, and the game business is a workforce of men that is paid more, and the two are usually spoken about as one industry.
What the spread does not mean
Within each group the tenure range is wide — 2.19 to 6.89 years in entertainment, 3.20 to 7.47 in games. It is tempting to read the low end as churn. The filing does not support that reading.
An average tenure figure moves for two opposite reasons: people leaving, and people arriving. A company that doubles its headcount in a year halves its average tenure without a single resignation. HYBE at 2.96 years and Krafton at 3.20 are both companies that grew hard. We cannot separate the two causes from one number a year, so we do not rank these companies and we do not call any of them a bad place to work.
What the numbers do support is the size of the thing. Fourteen companies. 15,072 people. That is the listed core of an industry whose output the world has now watched for 23.7 billion hours — and it is roughly the payroll of a single mid-sized Korean manufacturer.
Where these numbers come from
Sources
- Financial Supervisory Service (Korea), DART — Annual report employee and executive disclosures (직원 등의 현황), filing year 2025, for 2,862 listed companies · https://dart.fss.or.kr
- Netflix — Top 10 weekly global hours viewed (Tudum), 265 weeks from 2021-07-04 to 2026-07-26 · https://www.netflix.com/tudum/top10
- Wikidata — Country of origin (P495 = Q884) used to identify Korean titles in the Netflix lists · https://www.wikidata.org
What we checked
- The 14 companies are the listed entertainment and game firms in the DART filing set, split by their own reported industry code: eight file as film, video and audio production or broadcasting, six as publishing (which is where Korean game companies file). A fifteenth company that screens as an entertainment holding, Daea TI, files as electrical equipment — it makes rail signalling gear — and is left out
- Tenure, headcount, pay, female share and years since incorporation are read from the same filing, not mixed across years. Every figure quoted as a group average is weighted by headcount, so a 3,262-person studio and a 170-person label each count for their own staff rather than one company one vote
- The market baseline is the same field on the same filing for all 2,779 listed companies that report both tenure and headcount — 1,865,198 workers. It is the identical measure, so the comparison is like for like
- We tested whether these companies simply being younger explains the short tenure. It does not, on its own: across the 14, tenure and years since incorporation correlate at only 0.30, and the ratio of one to the other ranges from 13% to 46% with no clear pattern. The finding is reported as a negative
- The Netflix figures count Korean titles appearing in the weekly global Top 10 across 265 consecutive weeks. Hours viewed are as Netflix publishes them; we did not model or estimate any week
What we left out, and why
- Any claim that the 15,072 people made the 23.7 billion hours. The catalogue is not attributed to these 14 companies — many Korean titles come from private studios, and some of these companies' revenue has nothing to do with Netflix. The two numbers are placed side by side to size an industry, not to link one to the other
- Privately held agencies and studios, which file no comparable disclosure. Several of the busiest production houses in Korea are private, so this is the listed core of the industry and not the whole of it
- Any ranking of the companies against each other. A short average tenure at a young or fast-hiring company is not a worse company, and this data cannot separate people leaving from people newly arriving — a firm that doubles its headcount mechanically halves its average tenure
- Why tenure differs between any two named companies. We can show the spread; we cannot attribute it from a filing that reports one number per firm per year
- Pay comparisons against the wider market. The disclosed average pay per employee mixes full-year and part-year staff differently across filings, so the entertainment-versus-games gap quoted here is between these 14 only
The data behind this
Written from the same data
- Korea's content exports grew 63% in two years. Its viewing at the top of Netflix did not move.
- Korea's television exports moved to companies where people stay half as long
- Korean viewing is winner-take-all. Attention on the people in it is not.
- Netflix never says where a show is from. Every count of Korean content on it is an inference — including ours.