How long Korea’s listed-company CEOs have held the seat
Korean annual reports state, for each registered executive, how long that person has held the position. We read that field for 3,589 chief-executive seats across 2,769 listed companies — seats, not people, because a company can register more than one. The median is 9 years. The longest we are willing to print is 55 years. 378 filings — 10.5% of them — carried no figure we could read, and a further 147 stated a tenure longer than the company has existed. Neither group is filled in with a zero; both are counted and reported below.
The rows we set aside
147 seats report a tenure longer than their company has existed — a chief executive of fifty-plus years at a firm incorporated more recently than that. The usual explanation is mundane: a company is re-incorporated, renamed or spun out, the incorporation date resets, and the filed tenure carries over service at the earlier entity. We cannot tell which rows are that and which are an error, so we take them out of the distribution and print the count instead of deleting them. A further 49 seats sit at companies whose incorporation date we do not hold; those we leave in, because we have no basis to doubt them.
| Company | Ticker | Stated tenure | Company age |
|---|---|---|---|
| JinroDistillers | 018120 | 58 years | 41.8 years |
| DAESUNG INDUSTRIAL | 128820 | 56 years | 16.3 years |
| UNICK | 011320 | 54 years | 50.5 years |
| CHEIL GRINDING WHEEL IND | 001560 | 53 years | 51.8 years |
| DAESUNG INDUSTRIAL | 128820 | 52 years | 16.3 years |
| DAEBO MAGNETIC | 290670 | 51 years | 31.8 years |
Our own article on this data, Korea's listed CEOs: the median has run the company eleven years, set aside 148 companies on the same test. It counted one chief executive per company — the longest-serving where there are several — and reached a median of 11.0 years. This page counts every registered seat and reaches 9 years. Both numbers are right for what they count, and we print them side by side rather than quietly choosing one.
The distribution, not the average
An average would put a single number on 3,064 very different careers, and a single number read as a norm stops being a map and starts being a push. So here is the whole spread. The largest band is 10–20 years at 663 people, but it holds only 21.6% of the total — no band holds even a quarter.
| Tenure | CEOs | Share | Of those we could read |
|---|---|---|---|
| Under 1 year | 164 | 5.4% | |
| 1–3 years | 549 | 17.9% | |
| 3–5 years | 296 | 9.7% | |
| 5–10 years | 553 | 18.0% | |
| 10–20 years | 663 | 21.6% | |
| 20–30 years | 549 | 17.9% | |
| 30+ years | 290 | 9.5% |
Bands include the lower bound and exclude the upper one, so a CEO at exactly twenty years sits in the 20–30 row. Percentages are of the 3,064 tenures we could read, not of all 3,589 chief executives.
Both tails are crowded
The common shorthand is that Korean listed companies are owner-led and their chief executives stay for decades. 839 of them — 27.4% — have indeed held the seat twenty years or longer. But at the same time 713 (23.3%) have held it under three years. The quartiles say the same thing: 3 years at the first, 21 years at the third. This is not one population with a typical length. It is two groups sharing a table.
Full-time and part-time are not the same job
The filings mark whether the executive serves full-time. Among those we could measure, the full-time median is 10 years against 2 years for the 161 chief executives recorded as not full-time (140 of whom stated a readable tenure). That is a gap of years, not months — and it means any single median for "the Korean CEO" is averaging across two different arrangements.
The longest-serving
Ranked by stated tenure, not by company size. Ties are left in filing order; we do not break them.
| Company | Ticker | Chief executive | Tenure | Full-time |
|---|---|---|---|---|
| EehwaConstruction | 001840 | Choi Sam-gyu | 55 years | Yes |
| KYUNGNONG | 002100 | Lee Byeong-man | 55 years | Yes |
| PANGRIM | 003610 | Seo Jae-hui | 55 years | Yes |
| DONGIL INDUSTRIES | 004890 | Oh Sun-taek | 55 years | Yes |
| HeunguOil | 024060 | Kim Sang-u | 54 years | Yes |
| KOREA ELECTRIC TERMINAL | 025540 | Lee Chang-won | 53 years | Yes |
| UNION | 000910 | Lee Geon-yeong | 52 years | Yes |
| HAEIN | 003010 | Won Gyeong-hui | 52 years | Yes |
| HANSUNG ENTERPRISE | 003680 | Lim U-geun | 52 years | Yes |
| Wonpoong | 008370 | Lee Mun-hui | 51 years | Yes |
| MSC | 009780 | Kim Gil-je | 51 years | Yes |
| SJM HOLDINGS | 025530 | Kim Yong-ho | 51 years | Yes |
Romanised names are our transliteration of the Korean filing, and may differ from the spelling the person uses. Company names come from the regulator's own English register; where a ticker is missing from it we say so rather than inventing a name.
What this page does not tell you
- Whether long tenure is good or bad. Stability and entrenchment produce the same number. We counted years in the seat; we did not grade them.
- How Korea compares. We have not measured chief-executive tenure in any other market, so we do not put one here. A figure we did not count is not a figure we will print.
- Who holds seats at more than one company. The filings carry no birth date, so two identical names may be one person or two. We do not merge them, and we do not count concurrent appointments at all.
- What the 378 unmeasured CEOs look like. They may skew long or short. Leaving them out of the percentages is the honest option, not a neutral one, and we flag it rather than hide it.
- Which of the 147 set-aside rows are wrong. A re-incorporation and a bad join look identical from here. We removed them from the distribution because we cannot verify them, not because we know they are false.
- Anything about one company from this distribution. It is a shape across 2,769 firms. A shape is not a statement about any single seat.
Source: DART 전자공시 — 임원 현황 (2025 사업보고서). Tenure is the figure the company itself filed; we did not recompute it from appointment dates. Every number on this page is computed at build time fromceo-tenure.json by scripts/measure-ceo-tenure.mjs; none is transcribed by hand. Rows whose tenure we could not read are reported as unmeasured.