In FY2021, HD Korea Shipbuilding & Offshore Engineering Co., Ltd. (KRX:009540) reported revenue of ₩15,493.4bn and cost of revenue of ₩16,010.1bn. The ships cost more to build than they sold for, by ₩516.7bn, before a single won of overhead. Operating loss for that year was ₩1,384.8bn.
In the second quarter of 2026, the same group reported ₩8,927.0bn of revenue, ₩2,083.6bn of gross profit and ₩1,645.1bn of operating income. Operating margin: 18.4%.
The company attributes that improvement substantially to productivity. I think the contract book explains most of it, and the distinction matters because one is durable and the other has a known expiry date.
A US investor looking at the FY2021 numbers might assume something catastrophic happened in the yards. It didn't. What happened was accounting working exactly as designed.
Korean shipbuilders recognize revenue on long-term contracts by percentage of completion, and IFRS requires that when a contract is expected to be loss-making over its life, the entire expected loss is recognized immediately rather than spread across the build. Contracts signed in 2020 and early 2021, when newbuilding prices were near a decade low, ran headlong into a doubling of steel plate costs and a labour shortage. The moment those contracts turned onerous, the whole loss landed at once.
That is why FY2021 showed a ₩1.38tn operating loss on ships that were mostly still steel plate. And it is why the years since have looked so good: the losses were already taken. Ships delivering now consume provisions that were charged to a prior year's income statement, so their cost of revenue is artificially light while contracts signed at 2023 to 2025 prices carry genuine margin on top.
Cumulative operating losses across FY2018, FY2021 and FY2022 came to ₩2,221.8bn. Retained earnings fell from ₩16,180.6bn at the end of FY2019 to ₩14,300.4bn at the end of FY2022. They stood at ₩18,158.8bn at the end of June 2026. The company has earned back the hole and then some.
Test the productivity claim directly. Second-quarter 2026 revenue was ₩8,927.0bn against ₩7,428.4bn a year earlier, up 20.2%. Gross profit was ₩2,083.6bn against ₩1,302.0bn, up 60.0%.
If the improvement were productivity — more ship out of the same yard-hours — you would expect gross profit to grow somewhat faster than revenue, because fixed conversion costs get spread. Three times faster is a different phenomenon. That is the average selling price of the work in the revenue mix rising, or the average cost embedded in it falling because it was pre-provisioned, or both.
The annual series says the same thing at lower resolution. Gross margin ran negative 3.3% in FY2021, then 2.1%, 4.9%, 10.2% and 17.8% in FY2025. That is a straight-line climb over four years through periods of very different labour availability and yard utilization. Productivity does not improve in a straight line for four years. A backlog rolling from trough-priced contracts to peak-priced ones does.
None of this makes the earnings fake. Money earned on a well-priced contract is money. It does mean the improvement is a function of what was signed in 2023-25, and it will hold only as long as that vintage dominates deliveries.
If the story were operational efficiency, the overhead line should show it. It shows the opposite.
Selling, general and administrative expense was ₩765.8bn in FY2023 and ₩1,423.9bn in FY2025, an 86% increase. Revenue over the same two years went from ₩21,296.2bn to ₩29,933.2bn, up 41%. As a share of revenue, overhead rose from 3.6% to 4.8%, and it was 4.9% in the second quarter of 2026.
Some of that is straightforward: Korean industrial companies pay large performance bonuses in good years, and the shipbuilding unions negotiated hard after the loss years. Some is the cost of the engineering and sales organization needed to chase LNG and defence work. Neither is a problem. But a company whose overhead ratio rises by more than a point while it claims productivity gains is describing something narrower than the whole picture.
Take the other side seriously, because there is a real version of it.
The yards genuinely are more productive than in 2022, for reasons that have nothing to do with contract pricing. The workforce shortage that plagued Korean shipbuilding in 2022-23 eased as foreign worker quotas expanded. Learning-curve effects on repeat LNG carrier series are large, and HD Korea Shipbuilding has been building the same designs in volume — the company reported 16 LNG carrier orders in 2026 against seven the year before, and a series of identical hulls is the single biggest productivity lever a yard has.
Utilization is also as high as it has ever been. Order intake in the first half of 2026 reached $16.38bn, which the company put at 96.2% of its full-year target, against a backlog of more than 500 vessels and roughly three and a half years of work. A yard with no gaps in its dock schedule is a more efficient yard than one juggling.
So both things are true. My argument is only about proportion: when gross profit grows three times as fast as revenue, contract price is doing most of the work.
The obvious risk is the reverse of what happened in 2021. Contracts being signed today are being signed at prices set in a market where Chinese yards have taken share and where the newbuilding cycle has already run for three years. If steel or labour costs move against a book signed at today's prices, the same provisioning mechanic that produced the FY2021 loss produces another one, and it lands all at once rather than gradually.
The second risk is more mundane. The margin cannot keep climbing. Going from negative 3.3% gross margin to 23.3% in five years was a recovery from an artificially depressed base. The next five years start from a normal one.
The company's disclosure of backlog by delivery year and by contract vintage, which it provides in outline at results presentations. If the 2027 and 2028 delivery slots are filled predominantly with 2024-25 contracts, margin holds. If 2026 order intake at $16.38bn came in at prices materially below the 2024 book, the peak is visible from here.
The specific number to watch in the third quarter is gross margin against the second quarter's 23.3%. Two more quarters above 22% would suggest a plateau rather than a peak, and would make the productivity argument considerably more credible than the arithmetic currently does.
kstock reads DART every morning and writes up what moved — the contract, the buyback, the number that does not add up. The daily post and a Saturday roundup, by email.