On April 24 the board of Hyundai Mobis Co., Ltd. (KRX:012330) approved ₩500bn to buy its own shares on the open market and cancel every one of them. The filing did the arithmetic for investors: at the previous day's close of ₩442,500, that money would retire 1,129,943 shares.
Buying ran from April 27 to July 21. The amended filing reports the outcome. The company spent ₩499,999,909,000, which is within ₩91,000 of the budget, and acquired 911,998 shares.
That's 217,945 fewer shares than the plan implied, a shortfall of 19.3%. Divide the money by the shares and the average purchase price was ₩548,246.
The mechanism is what produced the outcome, and it is worth spelling out because Korean buyback disclosures almost always work this way.
Mobis committed an amount of money, not a number of shares. The board fixed ₩500bn and let the market decide how much stock that bought. Over the following three months the share price rose well above the ₩442,500 reference, and every won spent at the higher price retired less of the company. By the end the average execution was 23.9% above the price used to size the programme.
A share-based mandate would have inverted the risk. Committing to retire 1,129,943 shares regardless of price means spending more when the stock is expensive and less when it is cheap, which is worse in a rising market and better in a falling one. Neither approach is right in the abstract. What a won-denominated mandate does guarantee is that the company buys the most stock precisely when the stock is cheapest, and the least when it is dearest, which is the correct behaviour, and it also guarantees the announced share count is fiction if the price moves at all.
Nobody was misled. The April filing states plainly that the share count was calculated from the April 23 close and that the actual quantity could change with the price. This is disclosure working as designed. It's the outcome that's unflattering.
Against 90,732,583 common shares outstanding before cancellation, 911,998 shares is 1.005%. The 1,129,943 the reference price implied would have been 1.245%.
So roughly 24 basis points of share count reduction went missing. On a company capitalised at ₩40.55tn that's about ₩97bn of foregone value transfer to continuing holders, if you measure it that way. It is not a disaster. It is also not nothing, and it happened inside a single three-month window.
The cancellation itself completed on August 3, taking the common count to 89,820,585, which matches the exchange's figure at the August 27 close. Legal basis was the proviso to Article 343(1) of the Commercial Act, so the share count fell and share capital did not. The buying was brokered by Hyundai Motor Securities, a group affiliate.
Book value per share is the cleaner test for a company trading below book, and Mobis is.
Total equity at June 30 was ₩51.68tn. Across 89,820,585 shares that's ₩575,415 of book per share. The average purchase price of ₩548,246 is 0.953 times book. The August 27 close of ₩451,500 is 0.785 times.
Buying back stock below book is accretive to book value per share, so the programme did add value. It added considerably less than it would have at the price available in April, or at the price available now. A company retiring shares at 0.95 times book when the same shares can be had at 0.79 times has left most of the accretion on the table.
The contrast within the group is instructive. Hyundai Motor's board approved a cancellation on August 26 of shares whose carrying value, computed at average acquisition cost, was ₩476,385m against a market value of ₩789bn. Those shares were accumulated over a long period at prices well below current levels. Same group, same standards, opposite result, and the difference is that one company bought opportunistically over years and the other bought a fixed sum inside a quarter.
None of this is an argument that Mobis isn't serious about returning capital. The evidence says it is.
The April cancellation traces back to a mid-to-long-term shareholder return plan disclosed on November 19, 2024, filed as a corporate value enhancement plan under Korea's value-up framework, and reinforced by a shareholder value policy disclosed on January 28, 2026. Dividends paid rose from ₩406.2bn in FY2024 to ₩583.4bn in FY2025, up 43.6%. Share capital of ₩491,096m at a ₩5,000 par value implies 98,219,200 shares issued at some point against 89,820,585 outstanding now, so something close to 8.5% of the company has been retired over the years.
The company also held nine separate investor relations events between mid-June and late August, which is a lot of management time spent in front of shareholders for a business that used to be famous for not caring what they thought.
The steelman for the April programme is straightforward: management could not have known the price would rise 24% during execution, deploying capital on a schedule is better than waiting for a bottom that may not come, and a ₩500bn commitment announced and completed on time is worth more in credibility than ₩500bn spent perfectly and quietly. Buybacks executed on a calendar always look bad in hindsight when the price rises during the window, and good when it falls.
That defence is fair. It's also why the design question matters more than this one instance.
The size and denomination of the next tranche. If the following resolution again commits a won amount over a fixed window, the same thing will happen whenever the price moves, and shareholders should expect the announced share count to be an estimate rather than a target. If it commits to a share count, or spreads execution over a longer period, or adds a price condition, the company has learned something from this round.
The second marker is simply whether another programme comes at all before year-end. The stock now trades 17.6% below what Mobis was willing to pay for it three months ago. A company that believed its own shares were worth ₩548,246 in June and has ₩5.45tn of cash on the balance sheet has an obvious use for some of it. Whether it acts is the clearest available signal about how the board actually values this business.
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.