Something took roughly ₩1.5tn out of Celltrion, Inc.'s (KRX:068270) retained earnings in the second quarter of 2026. The line stood at ₩4,589.3bn at the end of March and ₩3,288.0bn at the end of June. The company earned ₩371.3bn in the quarter and paid ₩167.8bn of dividends. Roll that forward and about ₩1.5tn is unaccounted for by profit and distributions.
The answer is share cancellation. When a Korean company retires treasury stock, the cost of those shares comes out of retained earnings while share capital stays put. Celltrion has been doing this at scale, and the second quarter was the largest single instalment yet.
What makes it worth writing about is that the company was simultaneously doing the opposite.
Share capital rose from ₩239,361,656,000 at the end of FY2025 and ₩239,448,948,000 at the end of Q1 2026 to ₩250,553,370,000 at the end of Q2. That is an increase of about ₩11.1bn. Share capital only rises when shares are created. In the same quarter that ₩1.5tn of stock was retired, Celltrion issued new shares.
The mechanism is a 무상증자, a bonus issue. The company capitalizes reserves and hands existing shareholders additional shares pro rata, free. Every holder ends up with more shares and the identical proportional claim. Nothing is created. The pie is cut into more slices.
Korean retail investors have historically read bonus issues as a positive signal, and prices often move on the announcement. Celltrion has leaned into this. It ran one of roughly 8.49m shares in 2025 and announced a larger one for 2026 as part of a shareholder-value package unveiled in May. The ₩11.1bn increase in share capital implies something in the region of 11m new shares this year, if the par value is ₩1,000.
Now set that against the cancellations. Press reporting puts the retirements at roughly 3.43m shares in 2024 and 4.97m in 2025, with a three-year cumulative total near 18.56m shares, about 8.4% of issued stock, and a cumulative value approaching ₩2tn once the ₩100bn tranche completed in June is included.
Those figures come from Korean press rather than a filing I read, so hold them loosely. But taken at face value, roughly 18.6m shares retired over three years against roughly 19.7m issued through two bonus issues. On the share count that is a wash. On the cash, one of the two cost the company ₩2tn and the other cost nothing.
The honest version is that these do different jobs.
The cancellation is capital allocation. Real cash bought real shares and destroyed them, which raises every remaining holder's claim on future earnings. Nearly ₩2tn is a serious commitment for a company that earned ₩1,031.5bn in FY2025, and it is roughly half of what I calculate as Celltrion's tangible equity. This is not a gesture.
The bonus issue is liquidity and optics. A stock trading in the six figures in won terms has a high absolute price per share, and issuing more of them lowers it, which widens the retail base. In a market where retail flow moves prices and where Celltrion has a large, loyal and famously vocal shareholder register, that is a defensible thing to want. It is also not creating value, and management should not be allowed to present the two in the same slide as though they were.
The combination does have one real cost: it muddies the per-share arithmetic. Earnings per share falls mechanically on a bonus issue and rises on a cancellation. Anyone trying to read Celltrion's EPS trend across 2024 to 2026 has to adjust for both, in opposite directions, in overlapping periods. That is more work than shareholders should have to do.
The cancellations run against a balance sheet with an unusual shape. Intangible assets were ₩14,187.8bn at the end of Q2 2026 out of ₩23,703.2bn of total assets. That is 59.9%. Total equity was ₩18,048.2bn, so subtracting intangibles leaves tangible equity somewhere near ₩3.9tn.
The intangible arrived with the 2023 merger. Before it, at the end of FY2022, intangibles were ₩1,622.3bn against ₩5,891.7bn of assets, or 27.5%. The absorption of Celltrion Healthcare added roughly ₩11.7tn of intangible assets in a single year, which is what you get when a company buys its own distribution channel at a market-based valuation.
That balance has not shrunk. It was ₩13,290.7bn at the end of FY2023, ₩13,701.6bn at FY2024, ₩13,778.3bn at FY2025 and ₩14,187.8bn now. Some of the growth is capitalized development spend and some is currency translation on assets held abroad. Either way, the largest single item on Celltrion's balance sheet is one that is only tested for impairment, never marked, and it is growing.
This is not an accusation. Distribution rights and product approvals are genuine assets and a biosimilar developer that owns its own channel is worth more than one that doesn't. But it does mean that when the company spends ₩2tn buying back stock, it is spending against a book value that is three-quarters unmarkable, and any writedown would land on shareholders who have already had their cushion returned to them.
One more capital allocation item, and it is a strange one. On 25 August 2026 Celltrion filed a clarification disclosure about press reports of a plant in Yesan, South Chungcheong Province. The filing confirms that in November 2023 it signed an MOU with the province and county for a roughly ₩300bn biologics and materials plant by December 2028, and that in February 2025 it signed an investment agreement memorandum with the province, the county and the Chungnam Development Corporation for a ₩300bn drug-product facility. Details including timing and amount may change.
The interesting part is the filing history. This is the eighth such clarification. The company has re-disclosed the same undetermined status every six months since 1 December 2023, and the next re-disclosure is scheduled for 24 February 2027. Thirty-three months after the first newspaper report, a ₩300bn project still sits in the "not yet confirmed" bucket.
Meanwhile property, plant and equipment did move: ₩1,244.6bn at the end of FY2024, ₩1,702.6bn at FY2025 and ₩1,834.5bn at the end of Q2 2026. So capacity spending is happening somewhere. Just not, apparently, in a form the company is ready to commit to on the record.
The 2027 shareholder return announcement, usually made alongside full-year results. The specific thing to look for is whether the bonus issue repeats. Two consecutive years of doing both suggests the company sees them as complements. A third year, with the cancellation program running at ₩2tn cumulative, would say management thinks the free-share signal is worth as much as the cash — and I'd want to hear the argument.
Second, the impairment note in the FY2026 annual report. Fourteen trillion won of intangibles has never been tested against a year in which the biosimilar approval barrier fell and price competition intensified. The assumptions in that note are more consequential to reported book value than anything in the income statement.
Third, the February 2027 Yesan re-disclosure. Either it converts to a board resolution with a number attached, or it becomes the ninth deferral, and at that point the project should probably be treated as shelved regardless of what the MOU says.
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.