Three days ago I wrote that HD Hyundai Heavy Industries (329180) had disclosed three serious accidents in eighteen days and shut every plant it operates for four days. What has changed since is a fourth notice, and it changes the shape of the problem rather than adding to it.
The filing landed on 31 August. A worker was found collapsed in the boiler machine room of Hanuri Hall on 28 August. One dead, no injuries. Under "nature of the accident" the company wrote that the cause of death is unknown.
Then this, in the notes: whether the death qualifies as a serious accident under the Occupational Safety and Health Act is unclear as of the disclosure date, and the company will amend the filing if the investigation finds otherwise.
Read that again. This is a company disclosing a serious accident while saying it may not be one.
Hanuri Hall is a welfare building, not a fabrication shop. A boiler machine room is not where ships get built. Police and the Ministry of Employment and Labour are on site reviewing records, and no one has yet said the death had a workplace cause.
A yard that had not disclosed three deaths in the previous seven weeks would probably have waited for the investigation. This one filed within three days and reported to the labour ministry the same day the body was found.
That's the first thing worth noticing, and it isn't cynical to say so. Disclosure behaviour is a signal about the environment a company thinks it's operating in. HD Hyundai Heavy has concluded that the cost of appearing to withhold now exceeds the cost of over-reporting.
Here's the part that matters commercially.
On 30 August — two days after a death nobody has attributed to a workplace cause, and one day before it was disclosed — the Ulsan East branch of the Busan Regional Employment and Labour Office issued a partial work stoppage order covering all A-frame ladders in use anywhere on the site.
Every ladder. Not the boiler room. Not the building. The whole yard's stock of a category of equipment that has nothing to do with the incident that prompted it.
Set that beside the two orders already in force from July: gondolas at Ulsan from 19 July, and the entire panel factory assembly line at Gunsan from 25 July. Three separate partial stoppages, each broader in scope than the accident that triggered it, in six weeks.
The regulator has stopped responding to individual events and started responding to a pattern. Once that happens, the company's exposure is no longer a function of how careful it is on any given day. It's a function of how many more incidents occur anywhere in its footprint before the pattern is judged to have broken, and the fourth one shows that "anywhere" now includes a canteen boiler room.
The June quarter produced ₩6,332.2bn of revenue and ₩1,039.9bn of operating income, a 16.42% margin at the best point of this shipbuilding cycle.
Divide the quarter's revenue by 91 days and a full day of lost output at the group's production is worth about ₩69.6bn. The four-day company-wide halt at the end of July, which the filing scored against ₩17.58tn of affected operations, therefore cost something in the region of ₩278bn of revenue timing — though in shipbuilding, where revenue is recognised on progress against contracts with fixed delivery dates, a stoppage doesn't destroy revenue so much as push it right and add cost to catch up.
The ladder order is different in kind. Gondolas and a panel line are discrete. Ladders are everywhere, used by everyone, in every trade. A ban on all of them across a working shipyard doesn't stop a process, it slows a thousand of them, and the effect will show up as schedule slippage and overtime rather than as a disclosed production halt.
That is precisely why it won't be easy to see in the third-quarter numbers, and why I'd treat the absence of a visible hit as weak evidence.
Not much, is the honest answer. The stock closed at ₩449,500 on 31 August, which is 3.6% above its 52-week low of ₩434,000 and 39.7% below the ₩745,000 high. It fell 2.77% on 28 August, before any of this was known.
Meanwhile the order book has kept filling. Six single supply contract disclosures between 30 June and 23 August, including four LPG carriers for ₩515.4bn signed on 21 August. The yard is being paid to build ships faster than the regulator is stopping it from building them.
So the market is pricing shipbuilding and largely ignoring the safety record. I think that's defensible in the near term and complacent beyond it, for a reason that isn't in any of the numbers above.
Korea's Serious Accidents Punishment Act attaches criminal liability to management, not just corporate fines. Four deaths in seven weeks at a single group is the fact pattern prosecutors build cases on, and the company's own filings are the evidence file. Whether the fourth death is ultimately classified as work-related is close to irrelevant to that dynamic — three already are.
A prosecution doesn't stop a yard. What it does is occupy the people who run it, at a moment when the same executives are being asked by the Korea Exchange about acquiring US shipyards, and when a decade's worth of orders has to be delivered on fixed dates and fixed prices.
The fourth incident may genuinely be unrelated to work — a medical event in a building with no industrial activity. If the investigation says so, the count goes back to three, the ladder order gets lifted, and the pattern I've described is partly an artefact of a company disclosing defensively.
Second, the yard is enormous. HD Hyundai Heavy employs and contracts tens of thousands of people at Ulsan alone, and any workforce of that size will have deaths from causes that have nothing to do with what it makes. Rate matters more than count, and the filings give a count.
Third, none of the stoppages so far has produced a visible dent in output. Second-quarter revenue was up 52.7% year over year and margins hit a cycle high while two work stoppage orders were in force.
Whether the A-frame ladder order gets lifted, and when. A short order is a warning. One that runs into October is a supervision regime, and it will show up in third-quarter cost of revenue rather than in the revenue line.
Then the classification of the 28 August death. The company has committed to amending the filing if the investigation reclassifies it, so there will be a document either way.
And the thing I'd actually count: whether a fifth notice appears before the end of September. Four in fifty-two days is a pattern. Five would make the regulator's posture on 30 August look restrained rather than disproportionate, and it would put every subsequent stoppage order on a different footing.
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.