On July 22 the board of SK Square Co., Ltd. (KRX:402340) resolved to cancel 34,388 common shares. Not 34,388 thousand. Thirty-four thousand, three hundred and eighty-eight, out of 131,958,386 outstanding, retired on July 31. The filing values them at ₩43,088,164,000 based on the previous day's close, which works out to exactly ₩1,253,000 a share.
That is 0.026% of the share count. The company it was meant to help trades roughly ₩112tn below the market value of its main asset. You can run the ratio yourself and it does not improve on inspection.
The gap first, because it sets the scale of the problem. At the August 27 close SK Square traded at ₩1,068,000 across 131,923,998 shares, for a market capitalisation of ₩140.9tn. SK hynix Inc. (KRX:000660) closed at ₩1,730,000 across 730,492,365 shares. A roughly 20% holding in that is worth ₩252.8tn. Call the shortfall ₩112tn, or about $80bn at an approximate ₩1,400 to the dollar. That's gross of the substantial Korean tax SK Square would owe on any sale, so the real economic gap is narrower, but not by an order of magnitude.
Against that, the 2026 shareholder return plan runs to ₩110bn of buybacks and ₩200bn of cash dividends. Total ₩310bn. Divide by ₩112tn and you get 0.28%.
The July cancellation is the part of that plan that actually executed, and it came from shares bought under a March 25 board resolution whose purchase was already complete. Par value stayed at ₩100 and capital was unchanged, which is the standard structure: retire treasury stock out of distributable profit, shrink the share count, leave the capital account alone. The mechanics are fine. The magnitude is what it is.
To be fair to management, the ₩310bn figure was framed as 70.7% of ordinary dividend income. That is a defensible payout ratio for a holding company living off dividends received. It is also a confession, because it means the return is sized against the trickle of cash coming in rather than against the discount it is supposed to close. Piece two of this series covered why those two numbers are so far apart.
SK Square spent four years telling investors that the discount would narrow once the portfolio was cleaned up. That project is now essentially done.
On June 29 the company filed a holding-company subsidiary withdrawal for One Store Co., Ltd., the app marketplace it had held at 45.78% through 10,409,600 shares. The stake went to zero. Subsidiary count went from six to five. The filing puts SK Square's carrying value at ₩82,186m, or 1.24% of separate-basis total assets at the end of 2025, and gives the reason as simple disposal of shares. Press reports put the buyer as Nexusworth, a blockchain venture founded by a former Wemade chief executive, and the price at ₩62.6bn. The DART filing itself doesn't state a price, so that comparison rests on the press number rather than the disclosure. If it's right, SK Square took roughly ₩20bn less than book for an asset it once planned to float.
One Store followed a queue. 11Street moved to SK Planet and Incross to SK Networks in intra-group transfers, and SK Shieldus, NanoEnTech and Dreamus had gone earlier. You can see the cumulative effect in the top line without needing a single press release: revenue of ₩2.37tn in FY2022, ₩2.01tn in FY2023, ₩1.65tn in FY2024, ₩1.41tn in FY2025. Four consecutive years down, by design.
So the promise was delivered and the discount did not close. At the end of March, SK Square's disclosed net asset value was about ₩123tn with the hynix stake accounting for roughly ₩118tn of it, or 96%. Whatever the remaining five subsidiaries are worth, they are not the reason this stock trades where it does.
The contrast landed on August 19. SK hynix's board resolved to acquire ₩40tn of its own shares and cancel all of them, roughly 24.07 million shares or 3.3% of the count at the prior close of ₩1,662,000. Acquisition runs three months from August 20, with cancellation promised within a week or two of completion.
₩40tn at the subsidiary. ₩110bn at the parent. The ratio is about 360 to one.
There's a silver lining in that for SK Square holders, and it's the part most people miss. If SK Square doesn't sell into hynix's buyback, and there's no indication it will, then retiring 3.3% of hynix's shares lifts SK Square's ownership percentage passively. That partially undoes the dilution from July's Nasdaq offering at zero cost. A holding company can grow its stake by standing still, which is a nicer position than most.
It still doesn't touch the discount on SK Square's own shares. Only SK Square can do that, and it is doing it at a rate of 0.026% per resolution.
The generous reading, and I think it's substantially correct, is that SK Square is boxed in rather than complacent.
The obvious source of funds for a discount-closing buyback is selling hynix shares. Two things block that. The first is tax: the embedded gain on a stake carried since the 2021 spin-off is enormous, and Korean corporate tax on a disposal would consume a large slice of any proceeds. The second is structural. Korean fair-trade law requires a holding company to keep a minimum stake in a listed subsidiary, and local coverage after the Nasdaq offering described SK Square as running tight against a 20% line. I have not verified which threshold applies to a company converted in 2021, since the rules changed that year and grandfathering may apply, and the filings don't address it. If the floor is 20% and the stake is now roughly 20%, SK Square has no shares to sell at all.
That leaves borrowing, and SK Square has started. Non-current liabilities rose from ₩1.55tn to ₩2.93tn over the first half. The ₩5.89tn transfer of capital reserves into retained earnings this year points the same way: it doesn't create cash, it raises the legal ceiling on distributable profit so that a debt-funded return is permissible. Those are the moves of a company preparing to return more than its dividend income allows.
The risk in that path is straightforward. Leveraging a holding company against a memory stake works while memory is working. SK Square's reported results already show what the other half of the cycle looks like, with a ₩2.33tn operating loss in FY2023. A company that has doubled its long-term debt to fund distributions is a different animal going into a downcycle than one that hasn't.
The harder criticism of my own argument is that discount-closing buybacks rarely work anyway. Korean holding companies have traded below NAV for decades through every conceivable capital return programme, and a company that spent ₩5tn buying back stock at a 44% discount would still very likely trade at a discount afterward, having swapped a cheap asset for a temporary optical gain. On that view ₩310bn is not timid, it's disciplined.
The 2027 instalment of the three-year return policy, which should land with the FY2026 results early next year. The question is whether the number is derived from dividend income again or from the discount. A plan anchored to received dividends tells you the ceiling is structural and this stock will trade on hynix's price with a roughly constant haircut. A plan anchored to net asset value, funded by debt, tells you management has decided the gap is worth attacking.
The nearer marker is the hynix buyback completing around late November. Watch SK Square's disclosed stake in its Q3 report against the 20.5% it held before July. If the number has recovered toward 20.5% without SK Square spending anything, the passive mechanism is working. If it hasn't, something else moved, and the filings will have to say what.
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.