On 24 July 2026 the board of Woori Financial Group Inc. (KRX:316140) sat down and did three things in one session. It approved a quarterly cash dividend of ₩220 a share. It approved a trust contract to buy back stock. And it committed to cancelling ₩150bn of the shares that trust acquires. Seven outside directors attended, none were absent, and the disclosures went out the same afternoon alongside the first-half results.
Read the numbers rather than the announcement and a specific picture emerges. This is a real return package, structured with more care than most Korean financial holdings manage, and it is small relative to what Woori's peers have been doing. The reason is not reluctance. It is that Woori spent its spare capital buying two life insurers, and it is now returning cash out of what is left over.
The cancellation filing sets the arithmetic out plainly. ₩150bn at the ₩30,900 close of 23 July works out to 4,854,368 common shares. Total issued common shares are 728,089,682. So the program, if executed at that price, removes 0.67% of the company. The filing is careful to say the actual count will move with the share price and with how many shares the trust manages to buy between 24 July 2026 and 22 January 2027, with Korea Investment & Securities acting as broker.
Compare that to the dividend declared the same day. ₩220 per share across 728,035,737 dividend-eligible shares comes to ₩160.2bn — more than the buyback. A US investor used to bank holding companies where the repurchase dwarfs the dividend will find that inverted. Woori is still, overwhelmingly, a dividend story with a buyback attached.
The combined ₩310bn goes against ₩1,004.6bn of controlling-interest net income for the second quarter, per the 24 July disclosure. That is about 31%. Not stingy for a bank that is mid-integration. Not the 40%-plus that Korea's value-up program has been pushing the sector toward, either.
This part deserves explaining because it has no clean US analogue. The dividend disclosure states that the ₩220 is funded from amounts converted into retained earnings by reducing the company's capital reserve. Under Korea's Income Tax Act, a distribution sourced that way is treated as a return of capital rather than dividend income, so it is not taxable income for most recipients. The filing carves out an exception for certain large resident shareholders, where the exclusion is capped at the book value of the shares they hold.
Why this matters: Korea taxes dividend income heavily enough that it distorts payout behavior, and retail holders have long complained about it. Routing a distribution through a capital reserve reduction hands shareholders the same cash with a materially better after-tax outcome, at no cost to the company beyond a legal step at the general meeting level. Several Korean holdings have started doing this. Woori doing it on a recurring quarterly basis, rather than as a one-off, says something about how seriously it takes the retail shareholder base it inherited from its long stretch under government ownership.
There is a second piece of housekeeping in the filing that also reads as deliberate. The 10 August record date and 31 August payment date were fixed back on 6 February 2026, along with the record and payment dates for the first and third quarters. Korea only recently began letting companies declare the dividend before setting the record date, ending the old practice where investors bought blind. Woori pre-committing three quarters ahead is the strongest version of that reform in practice.
The obvious objection to a 0.67% buyback is that the company is capital-constrained. Look at the balance sheet and the constraint is less obvious than it should be.
Total equity was ₩40.9tn at the end of Q2 2026, up from ₩35.9tn at the end of FY2024 and ₩28.9tn at the end of FY2021. Retained earnings reached ₩29.4tn against ₩27.0tn eighteen months earlier. Share capital has been flat at ₩3,802.7bn since FY2023, so none of that equity growth came from issuing stock — it is retained profit and comprehensive income. Korean press coverage of the earnings call put the group's CET1 ratio around 13.7% at the second quarter, up roughly 0.11 percentage points from the prior quarter, though that is a preliminary figure from reporting rather than something I pulled from a filing.
That is the interesting tension. Woori consolidated something like ₩110tn of insurance assets over six quarters and its capital ratio went up rather than down. Insurance liabilities carry different risk weights than loans, and the group has evidently been managing risk-weighted assets tightly. So the small buyback is not a solvency story. It looks more like a company that wants a cushion before it commits to a bigger number, having just spent heavily and not yet proven the spending worked.
The cash flow statement supports the cautious read. Dividends paid ran ₩927.3bn in FY2025, ₩878.3bn in FY2024 and ₩978.4bn in FY2023 — a plateau, not a climb, and that consolidated line includes payments to minority holders of subsidiaries as well as to Woori's own shareholders. Against FY2025 net income of ₩3,227.5bn, the FY2025 figure is under 29%.
The most serious counter is that I'm reading a quarterly announcement as if it were a policy. It isn't. Korean banks have been ratcheting returns up in steps, and a ₩150bn tranche in July says nothing about what a February board meeting decides for the full year. If the insurance integration completes cleanly and CET1 holds near 13.7%, the natural next move is a larger program, and one Korean report has speculated about a repurchase in the ₩350bn range. That is speculation, not a filing, but it is not unreasonable speculation.
The second counter is that buyback size is the wrong metric for a stock trading around 0.55x reported book. Every won spent below book is accretive, so a small program executed steadily still compounds book value per share. And the cancellation structure here is the good kind: shares acquired through the trust get retired entirely, issued share count falls, and share capital stays put. No treasury overhang parked on the balance sheet for later reissue.
The risk running the other way is credit. Provisions for credit losses were ₩2,102.8bn in FY2025 against ₩1,716.3bn in FY2024, and the quarterly figure has only recently eased to ₩439.2bn in Q2 2026 from ₩526.8bn in Q1. Returns funded out of a benign provisioning quarter are the first thing to get pulled when the cycle turns. Woori's payout has been steady through worse quarters than this, so I don't think a cut is likely. Growth in the payout is what would stall.
The buying window closes 22 January 2027, and Woori will have to file the actual cancellation. That filing gives two hard facts: how many shares were really retired, and the average price paid. If the count comes in near 4.85m and the average price is close to ₩30,900, the program was executed as announced and the discussion moves to the size of the next one.
Before that, the February 2027 board meeting sets the annual return policy alongside FY2026 results. That is the number that matters. Watch whether the total announced there moves past 30% of controlling net income, and whether the capital-reserve mechanism gets extended to the year-end dividend or only applies to the quarterly ones. The second question is smaller but it tells you how much of the after-tax argument the company actually intends to make permanent.
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.